Alibaba / Qwen, released Apr 1, 2025
Qwen3 235B-A22Bprice, context, benchmarks and release details
Provisional: not enough results to rank yet Open weights 76% confidence 76 percent, Medium confidence, 4 of 8 expected sources in
50.3
SI Score
#107 of 142 ranked models
- Input, per 1M tokens
- $0.29Alibaba Model Studio pricingOfficial Alibaba Model Studio Global USD on-demand API; standard tier; Non-Thinking and Thinking modes. Output uses non-thinking rate when both modes are available; thinking-only products use their thinking rate. Cache, Batch, free quotas and regional rates excluded
Retrieved Oct 9, 2026 · factual citation
Open source ↗ - Output, per 1M tokens
- $1.15Alibaba Model Studio pricingOfficial Alibaba Model Studio Global USD on-demand API; standard tier; Non-Thinking and Thinking modes. Output uses non-thinking rate when both modes are available; thinking-only products use their thinking rate. Cache, Batch, free quotas and regional rates excluded
Retrieved Oct 9, 2026 · factual citation
Open source ↗ - Context window
- 131Kmodels.devPublished source fact
Retrieved Oct 9, 2026 · MIT
Open source ↗ - Max output
- 16.4Kmodels.devPublished source fact
Retrieved Oct 9, 2026 · MIT
Open source ↗ - Released
- Apr 1, 2025models.devPublished source fact
Retrieved Oct 9, 2026 · MIT
Open source ↗
How the score breaks down
Coding (weight 40 percent) 21.4
Math (weight 15 percent) 68.9
Preference (weight 15 percent) 69.7
Reasoning (weight 30 percent) 70.7
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, 05:13 UTC.
Around it on the leaderboard
- 105 GPT-5 Mini 50.6
- 106 Qwen3 32B 50.4
- 107 Qwen3 235B-A22B 50.3
- 108 Qwen3 Max 50.2
- 109 GPT-4o (2024-05-13) 49.4
Benchmark results
4 benchmarks, 5 resultsEach 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 70.7%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] Published Jun 3, 2025
Retrieved Oct 9, 2026 · CC-BY
Open source ↗ 70.7
Open source ↗ 70.7
MATH Level 5math 68.9%Epoch AI BenchmarkingEpoch-owned evaluation mean score; scores only, no benchmark questions [variant] Published Jun 3, 2025
Retrieved Oct 9, 2026 · CC-BY
Open source ↗ 68.9
Open source ↗ 68.9
SWE-bench Pro (public)coding 21.4%Official model cards via models.devLab-reported; metric resolve rate; transcribed by MIT models.dev catalog; not independently evaluated [variant] public
Retrieved Oct 9, 2026 · factual citation; MIT transcription
Open source ↗
21.4 2 settings
- Setting 1 21.4%Official model cards via models.devLab-reported; metric resolve rate; transcribed by MIT models.dev catalog; not independently evaluated [variant] public
Retrieved Oct 9, 2026 · factual citation; MIT transcription
Open source ↗ - Setting 2 21.4%SWE-bench Pro (public)Published steward score [variant] Published Jan 27, 2026
Retrieved Oct 9, 2026 · factual citation
Open source ↗
LMArena Textpreference 1366.3 eloLMArena / ArenaPublished source fact [variant] text / overallPublished Oct 8, 2026
Retrieved Oct 9, 2026 · CC-BY-4.0
Open source ↗ 69.7
Open source ↗ 69.7
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
- 60% of expected source weight
Reported (4)
- Epoch AI BenchmarkingOct 8, 2026
- Official model cards via models.devOct 8, 2026
- LMArena / ArenaOct 8, 2026
- SWE-bench Pro (public)Oct 8, 2026
Awaiting (4)
- ARC Prize11% of weight
- Humanity’s Last Exam11% of weight
- LiveBench11% of weight
- Terminal-Bench6% of weight
Confidence rises as pending sources publish. Some sources never cover some models, so confidence reaches 100% at 80% of expected weight.