Across 9 comparable benchmarks, Qwen3-Max-Thinking has better reported results on 8 and worse on 1.
Strong evidence9 of 9 shared results countedHow we compare
Price basis: Listed rates vs standard rates. The price section below shows the applicable periods.
| Fact | Qwen3-Max-Thinking | GPT OSS 120B |
|---|---|---|
| Context window | 1M tokens | 128K tokens |
| Input / 1M | — | $0.350 |
| Output / 1M | — | $0.750 |
| Blended (4:1) | — | $0.430 |
| Status | Deprecated | Available |
| Released | January 26, 2026 | August 6, 2025 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 5 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
Qwen3-Max-Thinking ahead on 2 of 3 counted
Qwen3-Max-Thinking ahead on 2 of 2 counted
Qwen3-Max-Thinking ahead on 1 of 1 counted
Qwen3-Max-Thinking ahead on 1 of 1 counted
No shared comparable results for Abstract reasoning, Mathematics, Research & search, Long context, Multimodal, Cybersecurity.
Published scores, with matching modes where available. Gaps show Qwen3-Max-Thinking relative to GPT OSS 120B. Reading the results
| Benchmark | Qwen3-Max-Thinking | GPT OSS 120B | Gap |
|---|---|---|---|
| LiveCodeBenchAlgorithmic Coding · Pass @K | 85.9Thinking · no tools | 87.8≠Thinking High · no tools | 0.98×#22 vs #14 of 113 |
| SWE-bench VerifiedRepository Engineering · Accuracy | 75.3Thinking · no tools | 60.1Thinking · no tools | +15.2 pts#36 vs #78 of 108 |
| Terminal Bench HardAgentic Development · Accuracy | 24.2Thinking · tools | 23.5≠Thinking High · tools | +0.7 pts#83 vs #88 of 157 |
| CritPtScientific Reasoning · Score | 1.7Thinking · no tools | 1.1≠Thinking High · no tools | 1.55×#92 vs #103 of 134 |
| GPQA DiamondScientific Reasoning · Accuracy | 87.4Thinking · no tools | 80.1Thinking · no tools | +7.3 pts#68 vs #111 of 200 |
| Pinch BenchTool Orchestration · Accuracy | 80.3Thinking · tools | 60.6Thinking · tools | +19.7 pts#24 vs #36 of 38 |
| IF BenchInstruction Following · Accuracy | 70.9Thinking · tools | 69.0≠Normal · no tools | +1.9 pts#24 vs #28 of 38 |
| HLEKnowledge Exams · Accuracy | 49.8Thinking · tools | 19.0Thinking · tools | +30.8 pts#51 vs #107 of 151 |
| MMLU-ProKnowledge Exams · Accuracy | 85.7Thinking · no tools | 79.0Thinking · no tools | +6.7 pts#27 vs #65 of 153 |
Algorithmic Coding · Pass @K
Repository Engineering · Accuracy
Agentic Development · Accuracy
Scientific Reasoning · Score
Scientific Reasoning · Accuracy
Tool Orchestration · Accuracy
Instruction Following · Accuracy
A dash means no result is available.
| Benchmark | Qwen3-Max-Thinking | GPT OSS 120B |
|---|---|---|
| AIME 2024 | — | 96.6 |
| AIME2025 | — | 97.9 |
| C-Eval | 93.7 | — |
| HLE-Verified | 38.2 | — |
| IMO-AnswerBench | 83.9 | — |
| MathArena Apex | — | 1.0 |
| MMLU | — | 90.0 |
| SciCode | — | 34.0 |
| Aider-Polyglot | — | 41.8 |
| APEX-SWE | — | 3.3 |
| CodeForces | — | 2622 |
| Vibe Code Bench v1.1 | 3.5 | — |
| APEX-Agents | — | 4.4 |
| PinchBench v2 | — | 44.8 |
| AA Intelligence Index (historical versions) | — | 24.1 |
| AA Intelligence Index v5.0.1 | — | 3.8 |
| ECI | — | 140 |
| LiveBench | — | 46.1 |
| Creative Writing | — | 956 |
| GDP.pdf | — | 4.0 |
| AutomationBench | — | 0.15 |
| Harvey Lab-AA | — | 13.9 |
| τ²-Bench | 82.1 | — |
| τ³-Banking | — | 12.8 |
| SimpleBench | — | 22.1 |
| Vending-Bench 2 | — | -21.5 |
| MedCode | 31.4 | — |
| MedScribe | 72.7 | — |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
Qwen3-Max-Thinking: Listed text rates; pricing tier not confirmed.
GPT OSS 120B: Standard pricing. Cerebras Shared Inference official pricing (gpt-oss-120b) verified 2026-11-13
| Price / 1M tokens | Qwen3-Max-Thinking | GPT OSS 120B | Ratio |
|---|---|---|---|
| Input | — | $0.35 | — |
| Output | — | $0.75 | — |
| Cached input | — | — | — |
| Cache write | — | — | — |
| Blended | — | $0.43 | — |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
| Benchmark | Qwen3-Max-Thinking | GPT OSS 120B |
|---|---|---|
| AA Intelligence Index v5.0.1 | Not reported | |
| AutomationBench | Not reported | $0.01 / task Average · Pass Rate: 0.15 Thinking High (With Tools) · zapier-private-heldout |
| MedCode | Not reported | |
| MedScribe | Not reported | |
| Vibe Code Bench v1.1 | Not reported |
| Attribute | Qwen3-Max-Thinking | GPT OSS 120B |
|---|---|---|
| Availability | Deprecated | Available |
| Provider | Alibaba | OpenAI |
| Model type | Reasoning model | Reasoning model |
| First release | January 26, 2026 | August 6, 2025 |
| Context window | 1M tokens | 128K tokens |
| Max output | 32K tokens | 128K tokens |
| Total parameters | 1.0T | 117B |
| Active parameters | — | 5.1B |
| Architecture | Dense | Mixture of Experts (MoE) |
| Documented capabilities | Reasoning, Multilingual | Reasoning, Multilingual |
| Knowledge cutoff | 2024-10 | — |
| Input modalities | Text | Text |
| Output modalities | Text | Text |
| Weights | Closed / API only | Open weights |
| Licence | — | Apache 2.0 |
| Commercial use | Proprietary | Commercial use allowed |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used Qwen3-Max-Thinking or GPT OSS 120B? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.