Skip to content
AI Atlas
ModelActiveOpen weightsIdentity probable

gpt-oss-120b

OpenAIfamily · gpt-ossdevelopers.openai.com/api/docs/models/gpt-oss-12

Most powerful open-weight model, fits into an H100 GPU

Open in Graph
data quality78

Updated 4 h ago · first seen 11 Sept 2026

model_01M293VC69K5D58EP5QCDWXVWY

Overview

Identity

Canonical model
Yesidentity confidence: mediumOne row per real model release. Artifacts (checkpoints, quantisations, conversions) and folded evaluation variants point here.
Official checkpoints
official_checkpoints = hf_repo identifiers carried by the model itself; artifacts are separate entities pointing here through canonical_id.
Artifacts
1 quantization0 official · 1 third-partySeparate entities (checkpoint · quantization · conversion · packaging) pointing to this model through canonical_id.
Provider deployments
5
API aliases
fireworks/gpt-oss-120bgpt-oss-120bgpt-oss-120b-lowopenai/gpt-oss-120bIdentifiers under which providers and evaluators refer to this model.
Folded evaluation variants
2Effort / thinking variants (…-high, …-non-reasoning) are result configurations of this model, not separate models. Their old URLs redirect here.

Openness

Open weights— weights downloadable under Apache-2.0; commercial use allowed; redistribution allowed; derivatives allowed; 4 dimensions unknown.

Weights downloadable under a permissive or Creative Commons licence allowing commercial use; code or data may be missing.

  • Weights

    Yes

  • Inference code

    —

  • Training code

    —

  • Training data

    —

  • Dataset

    —

  • Commercial use

    Yes

  • Redistribution

    Yes

  • Derivatives

    Yes

Licence: Apache License 2.0 (permissive · SPDX Apache-2.0 · stated as “apache-2.0”)

dimensions marked null are unknown, not false

Key facts

Release date

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 9 h agomedium

Knowledge cutoff

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Official page
developers.openai.com/api/docs/models/gpt-oss-120b

Source:OpenAI Platform docsT1observed 14 h agohigh

Model card
huggingface.co/openai/gpt-oss-120b

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 14 h agomedium

API model id

Source:OpenAI Platform docsT1observed 14 h agohigh

Openrouter id

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Architecture

Architecture

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Model type

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Parameters

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 14 h agomedium

Tokenizer

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Weights dtype

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

File size

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Library name

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Pipeline tag

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 14 h agomedium

Hugging Face repo

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Capabilities

Modalities

Modalities
text
Input
text
Output
text

Capabilities

  • Tool calling

    Yes

    OpenRouter public model & pricing listing · T2

  • Structured output

    Yes

    OpenRouter public model & pricing listing · T2

  • Reasoning

    Yes

    OpenRouter public model & pricing listing · T2

  • Vision

    Unavailable

  • Audio

    Unavailable

  • Fine-tuning available

    Unavailable

Context window

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Max output

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Knowledge cutoff

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Tokenizer

Source:OpenRouter public model & pricing listingT2observed 14 h agomedium

Benchmarks35

Compare with another model →

Comparable same task and conditions · Partially comparable same task, conditions differ (effort, temperature, judge) · Not comparable different variant or metric

Benchmark results grouped by comparability group
Benchmark · groupBest scoreTrustConfigurationResultsvs leaderEvaluatedSource
Terminal-Benchagentic · accuracy · variant=v2.1 · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisvariantv2.1reasoning_efforthighconditions differ across rows → partially comparable4non-primary groupobs. 12 Sept 2026artificialanalysis.aiT2
Terminal-Benchagentic · accuracy · variant=v4.0 · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisvariantv4.0reasoning_efforthighconditions differ across rows → partially comparable2non-primary groupobs. 12 Sept 2026artificialanalysis.aiT2
Terminal-Benchagentic · accuracy · variant=hard · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisvarianthardreasoning_efforthighconditions differ across rows → partially comparable4−42.4 ptvs gpt-5.6-solobs. 12 Sept 2026artificialanalysis.aiT2
τ²-benchagentic · pass^1 · variant=Telecom · evaluator=Artificial AnalysisIndependentevaluatorArtificial AnalysisvariantTelecomreasoning_efforthighconditions differ across rows → partially comparable2non-primary groupobs. 12 Sept 2026artificialanalysis.aiT2
τ²-benchagentic · pass^1 · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysis2−33.3 ptvs Z.ai GLM 5.2obs. 11 Sept 2026artificialanalysis.aiT2
Aider polyglotcoding · pass_rate_2Official boardedit_formatdiffreasoning_efforthighconditions differ across rows → partially comparable1−44.9 ptvs gpt-56 Aug 2025aider.chatT2
Aider polyglot — well-formed responsescoding · percent_cases_well_formedOfficial boardedit_formatdiffreasoning_efforthighconditions differ across rows → partially comparable1−20.9 ptvs Codestral 25.016 Aug 2025aider.chatT2
SciCodecoding · accuracy · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisreasoning_efforthighconditions differ across rows → partially comparable2−29.0 ptvs Claude Fable 5.1obs. 12 Sept 2026artificialanalysis.aiT2
SWE-bench Verifiedcoding · resolved · board=Verified · system=mini-SWE-agentOfficial boardboardVerifiedsystemmini-SWE-agent1−50.8 ptvs Claude Opus 4.57 Aug 2025swebench.comT2
Artificial Analysis Intelligence Indexcomposite · indexIndependentreasoning_efforthighversion4.3conditions differ across rows → partially comparable4−41.0vs Claude Fable 5.1obs. 12 Sept 2026artificialanalysis.aiT2
IFBenchinstruction-following · accuracy · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisreasoning_efforthighconditions differ across rows → partially comparable4−14.3 ptvs Grok 4.3obs. 12 Sept 2026artificialanalysis.aiT2
Humanity's Last Examknowledge · accuracy · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisreasoning_efforthighconditions differ across rows → partially comparable4−39.5 ptvs Claude Fable 5.1obs. 12 Sept 2026artificialanalysis.aiT2
GPQA Diamondreasoning · accuracy · variant=Diamond · evaluator=Artificial AnalysisIndependentevaluatorArtificial AnalysisvariantDiamondreasoning_efforthighconditions differ across rows → partially comparable2non-primary groupobs. 12 Sept 2026artificialanalysis.aiT2
GPQA Diamondreasoning · accuracy · variant=GPQA Diamond · evaluator=Artificial AnalysisIndependentevaluatorArtificial AnalysisvariantGPQA Diamond2−18.1 ptvs gpt-6-astraobs. 11 Sept 2026artificialanalysis.aiT2

Current rows only, grouped by benchmark → canonical metric → comparability group (task configuration). Effort variants folded into this model appear as rows of the same group. 35 current rows in total. “vs leader” compares with the current leader of the benchmark's primary group only; other groups are not directly comparable. Comparability rules →

Providers & Pricing7

All offers in the price terminal →
Provider deployments, cheapest output first
ProviderContextInput / 1MCached inOutput / 1MNative unitsStatusObservedSource
OpenAI APIcheapest outputopenai/gpt-oss-120b131.1Kout 118K——active5 h agosince 11 Sept 2026openrouter.aiT2
OpenRouteropenai/gpt-oss-120b131.1Kout 118K——active47 min agosince 12 Sept 2026openrouter.aiT2
Fireworks AIfireworks/gpt-oss-120b—$0.015—active7 h agosince 11 Sept 2026app.fireworks.aiT1
GroqCloudopenai/gpt-oss-120b131.1Kout 65.5K—output_tokens_per_second=500active1 h agosince 11 Sept 2026console.groq.comT1
OpenAI APIopenai/gpt-oss-120b:batch131.1Kout 118K——active5 h agosince 11 Sept 2026openrouter.aiT2
OpenRouteropenai/gpt-oss-120b:batch131.1Kout 118K——active47 min agosince 12 Sept 2026openrouter.aiT2
Together AIgpt-oss-120b———active13 h agosince 11 Sept 2026together.aiT1

USD per 1M tokens as published by each provider; native units (per-request fees, flex/priority tiers) are kept verbatim. Rows are append-only — every price change is kept in the history below. Cost of a workload →

Price history

Step lines per provider; amber markers are recorded changes. Click a marker or a row for the evidence behind that price.

Output price · USD / 1M tokens 5 providers

Output price history of gpt-oss-120b$0$0.20$0.40$0.60$0.80Sept 26Sept 26Sept 26Sept 26Sept 26Fireworks AI: first observed → $0.60 · 11 Sept 2026OpenAI API: first observed → $0.17 · 11 Sept 2026OpenAI API: $0.17 → $0.60 · 11 Sept 2026GroqCloud: first observed → $0.60 · 11 Sept 2026Together AI: first observed → $0.60 · 11 Sept 2026OpenRouter: first observed → $0.17 · 12 Sept 2026OpenRouter: $0.17 → $0.60 · 12 Sept 2026
  • Fireworks AI
  • OpenAI API
  • GroqCloud
  • Together AI
  • OpenRouter
  • OpenRouter$0.17 → $0.6012 Sept 2026
  • OpenRouterfirst observed $0.1712 Sept 2026
  • Together AIfirst observed $0.6011 Sept 2026
  • GroqCloudfirst observed $0.6011 Sept 2026
  • OpenAI API$0.17 → $0.6011 Sept 2026
  • OpenAI APIfirst observed $0.1711 Sept 2026
  • Fireworks AIfirst observed $0.6011 Sept 2026

Input price · USD / 1M tokens 5 providers

Input price history of gpt-oss-120b$0$0.05$0.10$0.15$0.20Sept 26Sept 26Sept 26Sept 26Sept 26Fireworks AI: first observed → $0.15 · 11 Sept 2026OpenAI API: first observed → $0.037 · 11 Sept 2026OpenAI API: $0.037 → $0.15 · 11 Sept 2026GroqCloud: first observed → $0.15 · 11 Sept 2026Together AI: first observed → $0.15 · 11 Sept 2026OpenRouter: first observed → $0.037 · 12 Sept 2026OpenRouter: $0.037 → $0.15 · 12 Sept 2026
  • Fireworks AI
  • OpenAI API
  • GroqCloud
  • Together AI
  • OpenRouter
  • OpenRouter$0.037 → $0.1512 Sept 2026
  • OpenRouterfirst observed $0.03712 Sept 2026
  • Together AIfirst observed $0.1511 Sept 2026
  • GroqCloudfirst observed $0.1511 Sept 2026
  • OpenAI API$0.037 → $0.1511 Sept 2026
  • OpenAI APIfirst observed $0.03711 Sept 2026
  • Fireworks AIfirst observed $0.1511 Sept 2026

Hardware fit37

Estimated

19 of 37 device × quantization combinations fit.

Run locally: your machine →
Estimated hardware fit
HardwareQuantizationDevice memoryEst. memoryFits
Apple M3 Ultra4bit—67.7 GB est.Yes
Mac Studio (Apple M5 Ultra)4bit—67.7 GB est.Yes
Apple M2 Ultra4bit—67.7 GB est.Yes
Apple M1 Ultra4bit—67.7 GB est.Yes
Apple M3 Max4bit—67.7 GB est.Yes
Apple M4 Max4bit—67.7 GB est.Yes
Mac Studio (Apple M5 Max)4bit—67.7 GB est.Yes
MacBook Pro (Apple M5 Max)4bit—67.7 GB est.Yes
Apple M2 Max4bit—67.7 GB est.Yes
NVIDIA A100 80GB4bit80 GB67.7 GB est.Yes
NVIDIA H100 SXM4bit80 GB67.7 GB est.Yes
NVIDIA H100 NVL4bit94 GB67.7 GB est.Yes
NVIDIA DGX Spark4bit128 GB67.7 GB est.Yes
NVIDIA H2004bit141 GB67.7 GB est.Yes
NVIDIA H200 NVL4bit141 GB67.7 GB est.Yes
NVIDIA B2004bit180 GB67.7 GB est.Yes
AMD Instinct MI300X4bit192 GB67.7 GB est.Yes
AMD Instinct MI325X4bit256 GB67.7 GB est.Yes
NVIDIA DGX B2004bit1,440 GB67.7 GB est.Yes
Apple M1 Max4bit—67.7 GB est.No
Apple M4 Pro4bit—67.7 GB est.No
Mac mini (Apple M5 Pro)4bit—67.7 GB est.No
MacBook Pro (Apple M5 Pro)4bit—67.7 GB est.No
Apple M3 Pro4bit—67.7 GB est.No
Apple M1 Pro4bit—67.7 GB est.No
Apple M2 Pro4bit—67.7 GB est.No
Apple M44bit—67.7 GB est.No
iMac (Apple M4)4bit—67.7 GB est.No
Mac mini (Apple M6)4bit—67.7 GB est.No
MacBook Air (Apple M5)4bit—67.7 GB est.No
MacBook Pro (Apple M5)4bit—67.7 GB est.No
Apple M24bit—67.7 GB est.No
Apple M34bit—67.7 GB est.No
Apple M14bit—67.7 GB est.No
NVIDIA GeForce RTX 30904bit24 GB67.7 GB est.No
NVIDIA GeForce RTX 40904bit24 GB67.7 GB est.No
NVIDIA GeForce RTX 50904bit32 GB67.7 GB est.No
Assumptions (7)
  • Estimated, not measured: weights = parameters × bytes/param × 1.15 runtime overhead (or the observed artifact file size when one is recorded).
  • bytes/param: 4bit = 0.5, 8bit = 1.0, fp16 = 2.0 (uniform quantization, no per-layer exceptions).
  • KV cache: 2 × layers × kv_heads × head_dim × 2 bytes × context × batch when the architecture is known; otherwise 0.5 GB per 8 192 tokens (× batch), independent of architecture (GQA/MLA models need less).
  • A model 'fits' when the estimate is at most the device memory minus 2 GB reserved for the OS and framework.
  • Mixture-of-experts models are estimated on total parameters (all experts must be resident); active parameters are ignored.
  • Device memory uses the largest configuration when several are listed (e.g. Apple silicon tiers).
  • Multi-GPU: device memories are summed; interconnect bandwidth, tensor-parallel replication and pipeline bubbles are not modelled.

Lineage

Open in Graph →
Explicit derived_from / fine_tuned_from / distilled_from relations stated by sources; artifacts collapsed by kind.
DESCENDANTS 0 · ARTIFACTS1 quantizationartifacts · collapsed1 quantization — artifacts · collapsedgpt-oss-120b116.8B params · this modelgpt-oss-120b — 116.8B params · this model

    Versions & Artifacts1

    Version history

    Context windowfirst observation only

    11 Sept 2026current

    Knowledge cutofffirst observation only

    11 Sept 2026current

    Licensefirst observation only

    11 Sept 2026current

    Max outputfirst observation only

    11 Sept 2026current

    Opennessfirst observation only

    11 Sept 2026current

    Parametersfirst observation only

    11 Sept 2026current

    Each hop is a claim: click a value for its source, tier and observation time. Nothing is overwritten — a new observation closes the previous claim.

    Artifacts 1

    quantization 1

    Papers1

    Timeline20

    Full timeline →

    Change history

    Temporal, append-only claims: a new observation closes the previous claim instead of overwriting it. Rewind the record with the as-of picker.
    2 claims · 1 propertiesShow all properties

    Aa median output tokens per secondmetric.aa_median_output_tokens_per_second2

    Claim history for Aa median output tokens per second
    ValueValid from → toStatusSourceConfidenceExtractor
    189.4 → currentcurrentArtificial AnalysisT2mediumdeterministic
    199.9 → currentcurrentArtificial AnalysisT2mediumdeterministic

    Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

    Provenance

    Attributed facts

    45

    Source tiers

    T1T25 / 40

    Freshest observation

    4 h ago

    Conflicts

    None

    Source documents 12

    Source documents
    SourceDocumentTypeTierLast observedSnapshots
    OpenAI Platform docsdevelopers.openai.com/api/docs/models.md model_docsT1· Official21 min ago1
    Fireworks AI — pricingdocs.fireworks.ai/serverless/pricing.md pricingT1· Official49 min ago2
    Groq — docs & pricingconsole.groq.com/docs/models model_docsT1· Official1 h ago5
    Together AI — pricingtogether.ai/pricing pricingT1· Official3 h ago1
    Hugging Face Hub (public pages, model cards, papers)huggingface.co/models?author=openai&p=0&sort=downloads listingT2· Quality secondary1 min ago8
    OpenRouter public model & pricing listingopenrouter.ai/api/v1/models catalogueT2· Quality secondary47 min ago11
    SWE-bench leaderboardsswebench.com/ leaderboardT2· Quality secondary4 h ago1
    Artificial Analysisartificialanalysis.ai/leaderboards/models leaderboardT2· Quality secondary4 h ago2
    Aider polyglot leaderboard (data file on GitHub)raw.githubusercontent.com/Aider-AI/aider/main/aider/website/_data/polyglot_leaderboard.yml leaderboardT2· Quality secondary4 h ago1
    Hugging Face Hub (public pages, model cards, papers)huggingface.co/openai/gpt-oss-120b/raw/main/README.md model_cardT2· Quality secondary7 h ago1
    Hugging Face Hub (public pages, model cards, papers)huggingface.co/openai/gpt-oss-120b model_pageT2· Quality secondary8 h ago5
    Hugging Face Hub (public pages, model cards, papers)huggingface.co/amd/gpt-oss-120b-w-mxfp4-a-fp8 model_pageT2· Quality secondary8 h ago5

    Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.

    Data quality (78/100) measures how well AI Atlas knows this entity — completeness, primary-source ratio, freshness, conflicts — never how good the model is. Methodology →