feat: add nvidia/deepseek-ai/deepseek-v4-pro-0813 by guillaumegay13 · Pull Request #353 · mnfst/modelparams.dev

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

guillaumegay13 · 2026-09-03T04:36:37Z

✨ What changed

Adds deepseek-ai/deepseek-v4-pro-0813 (nvidia) to the catalog, params cloned from nvidia/deepseek-v4-flash-0731.yaml.
Hosted serving: the model is developed by deepseek (OpenRouter: deepseek/deepseek-v4-pro-0813); this entry documents nvidia's serving of it, probed against nvidia's endpoint.

💭 Why

The provider serves it but the catalog doesn't list it. Param surfaces differ per serving provider, so each host gets its own probed entry (models.dev-style).

📝 Notes

Liveness ✅ only: the model answers on its real endpoint; unverified params are listed in the evidence table below.
Draft PR, auto-proposed by the modelparams agent.

Param verification evidence

param	check	ok	detail
`(baseline)`	accepted	✅	1-shot answers 200
`temperature`	min=0	✅
`temperature`	max=1	✅
`temperature`	range policed	❌	temperature=2 accepted — declared max 1 may be too low or unenforced
`top_p`	max=1	✅
`top_p`	range policed	✅	top_p=2 rejected
`max_tokens`	min=1	✅
`max_tokens`	max=16384	➖	too large for a non-streaming probe — upper bound from docs
`max_tokens`	range policed	❌	vendor reports max 1048576, YAML says 16384
`frequency_penalty`	min=-2	✅
`frequency_penalty`	max=2	✅
`frequency_penalty`	range policed	✅	frequency_penalty=4 rejected
`presence_penalty`	min=-2	✅
`presence_penalty`	max=2	✅
`presence_penalty`	range policed	✅	presence_penalty=4 rejected
`seed`	min=0	✅
`seed`	max=18446744073709552000	❌	Failed to deserialize the JSON body into the target type: invalid type: floating point `1.8446744073709552e+19`, expected i64 at line 1 column 136
`seed`	range policed	✅	seed=36893488147419103000 rejected
`stop`	declared	➖	type string has no probe — value carried from sibling/docs, not verified

vercel · 2026-09-03T04:36:41Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Sep 3, 2026 8:04am UTC

feat: add nvidia/deepseek-ai/deepseek-v4-pro-0813
19350ea

guillaumegay13 added the liveness-verified Model verified live on its endpoint; params sibling-cloned, not individually probed label Sep 3, 2026

github-actionsBot added the model Add a model that's missing label Sep 3, 2026

vercelBotdeployed to Preview September 3, 2026 04:37 View deployment

Merge branch 'main' into mp/add-nvidia-deepseek-ai-deepseek-v4-pro-0813
5b008e3

vercelBotdeployed to Preview September 3, 2026 08:04 View deployment

guillaumegay13 merged commit 6772626 into mainSep 3, 2026
13 checks passed