feat(claude): add Claude Fable 5.1

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.

[juliusmarminge]

Claude Fable 5.1 is available in Claude Code 2.1.257, but T3 Code does not yet expose it.

Add claude-fable-5-1 as the newest built-in Claude model with Fable effort and context controls, gated to Claude Code 2.1.257+. The model manifest is intentionally unchanged for the follow-up work. Usage pricing continues to come from LiteLLM.

Sources:

• https://platform.claude.com/docs/en/models/fable-5-1/overview
• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

Tests: vp test run apps/server/src/provider/Layers/ProviderRegistry.test.ts apps/server/src/usage/usagePricing.test.ts apps/server/src/usage/UsageService.test.ts
Typecheck: vp run --filter t3 typecheck

Model: gpt-5.6-sol
Harness: Codex

---

Note

Low Risk
Catalog and version-gating changes for a new model slug; the main behavioral shift is reordering upgrade messages when Fable 5.1 is unsupported.

Overview
Adds claude-fable-5-1 to the built-in Claude model catalog with the same Fable-style Reasoning effort ladder (including ultracode/ultrathink) and Context Window options (200k / 1M default).

The model is only listed when Claude Code is ≥ 2.1.257 (supportsClaudeFable51); older installs omit it and surface a new Fable 5.1 upgrade message as the first step in the existing version fallback chain in checkClaudeProviderStatus. normalizeClaudeCliEffort treats Fable 5.1 like other recent Fable/Opus models so xhigh is not downgraded to max.

Provider registry tests cover Fable 5.1 visibility at 2.1.257 vs 2.1.256 and the expected upgrade copy.

^{Reviewed by Cursor Bugbot for commit df0dcb8. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add claude-fable-5-1 model to Claude provider catalog

• Adds the claude-fable-5-1 built-in model entry with effort and contextWindow select options to the CLAUDE_MODEL_CATALOG.
• Gates the model behind MINIMUM_CLAUDE_FABLE_5_1_VERSION ("2.1.257") via the new supportsClaudeFable51 helper in getBuiltInClaudeModelsForVersion.
• Fixes normalizeClaudeCliEffort so xhigh is not coerced to max for claude-fable-5-1.
• Moves the top version-upgrade gate in checkClaudeProviderStatus from supportsClaudeOpus5 to supportsClaudeFable51, with fallback to older tier prompts.
• Risk: checkClaudeProviderStatus now checks supportsClaudeFable51 before supportsClaudeOpus5; any caller relying on the Opus 5 upgrade message as the top-level gate will see the Fable 5.1 message instead on older installs.

^{Macroscope summarized df0dcb8.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.2 KiB	13.3 KiB	+50 B (+0.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.3 KiB	6.4 KiB	+45 B (+0.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	+44 B (+0.1%)	66.4 KiB	✅
Codex	Live turn messages	8	9	+1 (+12.5%)	21	✅
Claude	Total thread wire	13.5 KiB	13.3 KiB	−162 B (−1.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−8 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.6 KiB	6.4 KiB	−154 B (−2.3%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.8 KiB	56.4 KiB	−1.4 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 2d156a8 · PR result: df0dcb8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This adds a user-facing Claude model and changes its provider’s effective default selection on supported installations, along with model-routing and upgrade-message behavior. The product-default change and runtime scope warrant human review.

Notes:

• No code objects were reviewed. Approvability was decided on eligibility alone.

^{You can add or adjust custom eligibility rules. Learn more.}

[q1]

Heads-up from #9078 (opened a few minutes after this one with the same catalog hunk, sorry for the overlap): with the catalog entry alone, claude-fable-5-1 renders under Legacy models in the picker. ClaudeDriver runs applyModelManifest on every snapshot, and the claudeAgent list in model-manifest.json does not name the slug, so isLegacyModel returns true until the manifest on main is updated. Might be worth adding it to the bundled manifest here (leaving claude-fable-5 current until a release ships the entry, since the remote manifest propagates immediately).

#9078 carries that manifest line plus fable / fable-5.1 / claude-fable-5.1 slug aliases and an adapter xhigh passthrough test; I'll rebase it down to those once this lands, or close it if you'd rather fold them in.

FWIW, verified end to end on Claude Code 2.1.257 (Claude Max): a thread on claude-fable-5-1 with contextWindow: 1m completes and reports a 1M window, and Claude Code strips the [1m] suffix for this model, so resolveClaudeApiModelId needs no change. Pricing: LiteLLM has BerriAI/litellm#39148 and BerriAI/litellm#39151 open for the claude-fable-5-1 rate, after which usage cost prices itself.