Skip to content
ModelCap

gpt-oss-120b

Current board #60Arena #188Open weights

OpenAI·openai/gpt-oss-120b

gpt-oss-120b is a language model published by OpenAI. 19 OpenRouter endpoints are published in the latest snapshot. It accepts text input and returns text. The published context limit is 131K tokens. The catalogue declares support for reasoning, tool use, structured outputs.

ModelCap Index · Current board
48.7#60
Measured · 86% support · score interval 48.7–54.9
Market Gravity
60.8
Secondary market signal
Input
$0.037
per 1M tokens
Output
$0.17
per 1M tokens
Context
131K
tokens
Providers
19
endpoints
Gateway spend
—
latest · Vercel share

Providers

Uptime measured over the last 30 minutes
ProviderInputOutputContextQuantUptime
CoreWeavefp4Cheapest$0.03$0.17131Kfp499.41%
DeepInfrabf16$0.037$0.17131Kbf16100.0%
Novitafp4$0.05$0.25131Kfp499.94%
Googleglobal$0.09$0.36131K—91.67%
DigitalOcean$0.055$0.385128K—100.0%
SiliconFlowfp8$0.05$0.45131Kfp8100.0%
Mancer 2fp8$0.09$0.50131Kfp899.50%
BaseTenfp4$0.10$0.50128Kfp4100.0%
Amazon Bedrock$0.15$0.60131K—100.0%
DeepInfraturbo$0.15$0.60131Kbf1699.53%
Together$0.15$0.60131K—100.0%
Nebiusfp4$0.15$0.60131Kfp499.87%
Amazon Bedrockeu-west-1$0.15$0.60131K——
Phala$0.15$0.60131K—99.84%
Groq$0.15$0.60131K—100.0%
Parasailfp4$0.10$0.75131Kfp487.09%
Mara$0.15$0.75131K—99.34%
Cerebrasfp16$0.35$0.75131Kfp1699.09%
SambaNova$0.14$0.95131K—94.28%

Hugging Face Inference Providers

Exact repository identity

Provider economics and telemetry published by the Hugging Face Router. These are operational signals, not benchmark results.

ProviderInputOutputContextTTFTThroughputFeatures
deepinfra
live
$0.037$0.17131K714 ms42.6 tok/stools · structured
novita
live
$0.05$0.25131K480 ms110.8 tok/stools · structured
nscale
live
$0.10$0.40131K578 ms107.2 tok/stools · structured
ovhcloud
live
$0.09$0.47131K350 ms90.5 tok/stools · structured
baseten
live
$0.10$0.50128K290 ms132.9 tok/stools · structured
fireworks-ai
live
$0.15$0.60131K303 ms217.0 tok/stools · structured
together
live
$0.15$0.60131K243 ms48.1 tok/stools · structured
scaleway
live
$0.171$0.684—377 ms136.5 tok/stools · structured
cerebras
live
$0.35$0.75131K259 ms1056.4 tok/stools · structured
groq
live
$0.15$0.75131K263 ms434.6 tok/stools · structured
featherless-ai
live
——————

Reported evaluation leads (20)

Excluded from score

Structured rows reported in this exact model repository at the pinned revision below. They are useful research leads, but remain quarantined until dataset revision, harness, configuration and identity pass ModelCap’s independent admission review.

Dataset / taskMetricValueRevisionProvenance
cais/hle
hle
—11.3
Unpinned
not published
Reported
Quarantined
cais/hle
hle
—8.6
Unpinned
not published
Reported
Quarantined
cais/hle
hle
—9.1
Unpinned
not published
Reported
Quarantined
cais/hle
hle
—5.2
Unpinned
not published
Reported
Quarantined
cais/hle
hle
—19
Unpinned
not published
Reported
Quarantined
cais/hle
hle
—14.9
Unpinned
not published
Reported
Quarantined
Idavidrein/gpqa
diamond
—80.8081
Unpinned
not published
Reported
Quarantined
Idavidrein/gpqa
diamond
—73.5
Unpinned
not published
Reported
Quarantined
Idavidrein/gpqa
diamond
—73.1
Unpinned
not published
Reported
Quarantined
Idavidrein/gpqa
diamond
—68.1
Unpinned
not published
Reported
Quarantined
Idavidrein/gpqa
diamond
—67.1
Unpinned
not published
Reported
Quarantined
Idavidrein/gpqa
diamond
—80.9
Unpinned
not published
Reported
Quarantined
Idavidrein/gpqa
diamond
—80.1
Unpinned
not published
Reported
Quarantined
LEXam-Benchmark/LEXam
open_question
—51.74
Unpinned
not published
Reported
Quarantined
LEXam-Benchmark/LEXam
mcq_4_choices
—47.71
Unpinned
not published
Reported
Quarantined
ScaleAI/SWE-bench_Pro
SWE_Bench_Pro
—16.2
Unpinned
not published
Reported
Quarantined
SWE-bench/SWE-bench_Verified
swe_bench_%_resolved
—52.6
Unpinned
not published
Reported
Quarantined
SWE-bench/SWE-bench_Verified
swe_bench_%_resolved
—47.9
Unpinned
not published
Reported
Quarantined
SWE-bench/SWE-bench_Verified
swe_bench_%_resolved
—62.4
Unpinned
not published
Reported
Quarantined
TIGER-Lab/MMLU-Pro
mmlu_pro
—80.8
Unpinned
not published
Reported
Quarantined

ModelCap Index · Current board

Methodology
48.7
Measured · 3 public benchmark observations across 3 boards
Current board
#60
Public rank basis
Measured (measured)
Index support
85.7% · strong
Index interval
48.7–54.9
Rank posterior
Not available
Top 5 / Top 10 probability
Not available / Not available
Posterior as of
Snapshot timestamp unavailable
Identity binding
Exact catalogue product · openai/openai/gpt-oss-120b · aggregates disclosed benchmark configurations · endpoint configuration not claimed · artifact metadata openai/gpt-oss-120b@b5c939de8f754692c1647ca79fbf85e8c1e70f8a (not the evaluation revision)
Direct benchmark observations
3
Breadth-adjusted BES point estimate
51.8
Bayesian Evidence Score · measured evidence input only

BES normalizes admitted public benchmark evidence for measured Index rows. Its legacy score and observables prior do not define the public language rank.

Observed capability
51.8
Evidence status
confirmed · 84% mass
General preference
55.2
Coding
38.3
Agents & tools
—
Reasoning
—
Evidence breadth
92%
Evidence coverage
86%
Market and catalogue signals · secondary, never the language Index rank
Usage (OpenRouter popularity)
42.8
Liquidity (providers × uptime)
94.2
Open reach (HF downloads)
83.1
Surface (context / tools / modalities)
70.2
Freshness
24.7
  • · OpenRouter popularity #26
  • · 17 live providers
  • · HF 30d downloads 4138863
  • · Catalogue: context 131072, reasoning, tools
  • · First seen 2025-08-05

These adoption and deployment observations remain context only. They do not change this model's ModelCap Index score or rank.

BES evidence selected · measured input only

The bounded 0–100 capability score blends each source’s competitive placement with its published achievement, then combines capability families. Evidence breadth adds a modest uncertainty adjustment; price and popularity are not part of benchmark-led rank.

Deployment readiness

Metadata index
88.5
strong · 94% metadata coverage
Not capability
Artifact reproducibility
85.0
Access & legal clarity
100.0
Deployability
99.0
Evaluation provenance
66.4

Missing: artifactReproducibility.base-lineage, evaluationProvenance.evaluation-dataset-revisions.

A metadata completeness and deployability index—not a safety certification, quality grade, or production-readiness claim.

Market activity

Venue coverage
Market Gravity
60.8
OpenRouter weekly popularity
#26 of 298

Market Gravity uses OpenRouter’s full-catalogue popularity order. Sparse Vercel top-ten observations appear only when published and do not affect the score.

Arena preference

Status
Ranked
Source
Arena
Official rank
#188 of 385
Rating
1352.5
95% confidence interval
1348–1357
Votes
31K
Source snapshot
6 Aug 2026, 13:00 UTC
Category
overall
Identity match confidence
100%

Market Gravity breakdown

Usage55%
42.8

OpenRouter weekly popularity across the full model catalogue.

Liquidity25%
94.2

Independent providers versus the model's open or closed cohort, weighted by uptime.

Open reach15%
83.1

Hugging Face 30-day downloads for open models; neutral for closed models.

Freshness5%
24.7

Time since first public availability, on a six-month half-life.

Specification

Context window
131K tokens
Max output
131K tokens
Inputs
Text
Outputs
Text
Tokenizer
GPT
Cached input tokens
Not offered
First seen on OpenRouter
5 Aug 2025
Knowledge cutoff
2024-06-30

Capabilities

  • Supported: Reasoning
  • Supported: Tool use
  • Supported: Structured output
  • Supported: Response format
  • Not supported: Moderated

Weights & access

Open weights. Downloadable weights are published under an OSI or free-culture license in the current repository metadata. This describes weight availability, not whether the full training stack qualifies as Open Source AI.

Access
Open weights
Licence
apache-2.0
Pinned revision
b5c939de8f75
Parameters
116.8B
Architecture
GptOssForCausalLM
Model type
gpt_oss
Hugging Face downloads (30d)
4.1M
Downloads (all time)
49.0M
Likes
5K
Repository
openai/gpt-oss-120b
Repository updated
26 Aug 2025, 17:25 UTC