OpenAI, released Aug 5, 2025

GPT OSS 120Bprice, context, benchmarks and release details

Provisional: not enough results to rank yet Open weights 79% confidence 79 percent, Medium confidence, 5 of 10 expected sources in
54.8
SI Score
#90 of 142 ranked models
Input, per 1M tokens
not yet reported
Output, per 1M tokens
not yet reported
Context window
131Kmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Max output
32.8Kmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Released
Aug 5, 2025models.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗

How the score breaks down

Coding (weight 40 percent) 28.0
Math (weight 15 percent) 88.9
Preference (weight 15 percent) 69.6
Reasoning (weight 30 percent) 75.8

Weights: reasoning 30%, math 15%, coding 40%, preference 15%. Results use fixed 0–100 scales before averaging, and thin evidence is pulled toward 50. Method si-v3-retained-evidence-2, computed Oct 9, 2026, 06:15 UTC.

Around it on the leaderboard

  1. 88 o3 55.4
  2. 89 Gemini 3.5 Flash Lite 55.2
  3. 90 GPT OSS 120B 54.8
  4. 91 Gemini 3 Flash Preview 54.8
  5. 92 Grok 4.3 54.6

Full leaderboard

Benchmark results

6 benchmarks, 6 results

Each row shows the best published result. Where a model was tested at several settings, such as reasoning effort, open the row to see each one. Hover or tap a value for its source.

GPQA Diamondreasoning 75.8%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] highPublished Dec 11, 2025 Retrieved Oct 9, 2026 · CC-BY
Open source ↗ 75.8
OTIS Mock AIME 2024–2025math 88.9%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] highPublished Dec 11, 2025 Retrieved Oct 9, 2026 · CC-BY
Open source ↗ 88.9
Aider Polyglotcoding 41.8%Aider polyglotPublished source fact [variant] Aider polyglot; 225 cases; 2 attemptsPublished Aug 6, 2025 Retrieved Oct 9, 2026 · Apache-2.0
Open source ↗ 41.8
SWE-bench Pro (public)coding 16.2%SWE-bench Pro (public)Published steward score [variant] Published Jan 27, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 16.2
SWE-bench Verifiedcoding 26%SWE-bench VerifiedSWE-bench published model plus agent result; harness retained, not a base model evaluation [variant] mini-SWE-agent; 1.7.0Published Aug 7, 2025 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 26.0
LMArena Textpreference 1365.4 eloLMArena / ArenaPublished source fact [variant] text / overallPublished Oct 8, 2026 Retrieved Oct 9, 2026 · CC-BY-4.0
Open source ↗ 69.6

Normalization uses a fixed 0–100 scale for each unit, independent of other models. Compare evaluation conditions before reading a small gap as decisive. “Lab-reported” marks the provider's own published figure.

Details and sources

Open weights
Yesmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
License
not yet reported
Input modalities
textmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
First seen by SuperIndex
Oct 8, 2026
Coverage
63% of expected source weight

Reported (5)

  • Aider polyglotOct 8, 2026
  • Epoch AI BenchmarkingOct 8, 2026
  • LMArena / ArenaOct 8, 2026
  • SWE-bench VerifiedOct 8, 2026
  • SWE-bench Pro (public)Oct 8, 2026

Awaiting (5)

  • ARC Prize10% of weight
  • Humanity’s Last Exam10% of weight
  • Official model cards via models.dev3% of weight
  • LiveBench10% of weight
  • Terminal-Bench5% of weight

Confidence rises as pending sources publish. Some sources never cover some models, so confidence reaches 100% at 80% of expected weight.

What changed

All releases · RSS feed

Alerts on this device

What to be alerted about
RSS feed