GPT-5.3-Codex vs Qwen3.6 Max
Full profiles: GPT-5.3-Codex (OpenAI) and Qwen3.6 Max (Alibaba): every benchmark, price and rank.
Updated · Data from 6 sources
As of 27 Oct 2026, GPT-5.3-Codex (xhigh) and Qwen3.6 Max (max) are neck and neck on quality (63.6 vs 62.2 on the BenchLeader Index, inside the ±5.7 margin of error), and Qwen3.6 Max (max) costs 1.6× less, which makes it the better buy for most workloads. Pick GPT-5.3-Codex for reasoning (68 vs 58); pick Qwen3.6 Max for instruction following (75 vs 74).
GPT-5.3-Codex benchmarks 
Qwen3.6 Max benchmarks Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | GPT-5.3-Codex | 61 vs 57 category score; SWE-bench (Vals) 78.0% vs 72.8% | Medium |
| Agents & tool use | GPT-5.3-Codex | 80 vs 72 category score; Terminal-Bench Hard 53.0% vs 43.9% | High |
| Reasoning | GPT-5.3-Codex | 68 vs 58 category score; CritPt 16.9% vs 3.7% | High |
| Knowledge | GPT-5.3-Codex | 66 vs 62 category score; AA-Omniscience 10.9 vs 9.2 | Medium |
| Long documents | GPT-5.3-Codex | 67 vs 65 category score; AA-LCR 83.3% vs 80.7% | Medium |
| Following instructions | Qwen3.6 Max | 75 vs 74 category score; IFBench 76.6% vs 75.4% | Medium |
| Chat quality (human preference) | – | Only Qwen3.6 Max has an Arena rating (65 category score) | Not comparable |
| Lowest cost | Qwen3.6 Max | $2.92 vs $4.81 per 1M blended; agentic coding session $0.109 vs $0.161 | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- GPT-5.3-Codex (xhigh) and Qwen3.6 Max (max) are neck and neck on quality (63.6 vs 62.2, within the ±5.7 margin).
- GPT-5.3-Codex (xhigh) is stronger in agents & tools, coding, composite, knowledge, long context, multimodal, reasoning.
- Qwen3.6 Max (max) is stronger in instruction following, human preference, maths.
- Qwen3.6 Max (max) is 1.6× cheaper ($2.92 vs $4.81 per 1M blended).
At a glance
| Spec | GPT-5.3-Codex | Qwen3.6 Max |
|---|---|---|
| Developer | OpenAI | Alibaba |
| Released | 5 Feb 2026 | 20 Apr 2026 |
| Context window (tokens) | 400,000 | 262,144 |
| Compared at | GPT-5.3-Codex (xhigh) | Qwen3.6 Max (max) |
| Open weights | No | No |
| Price in / out / cached input per 1M | $1.75 / $14.00 / $0.175 | $1.30 / $7.80 / $0.130 |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | GPT-5.3-Codex (xhigh) | Qwen3.6 Max (max) | Leader |
|---|---|---|---|
| Reasoning | 68.1 | 58.1 | GPT-5.3-Codex |
| Coding | 60.8 | 56.6 | GPT-5.3-Codex |
| Agents & tools | 80.1 | 72.1 | GPT-5.3-Codex |
| Maths | – | 64.9 | not comparable |
| Knowledge | 66.0 | 62.1 | GPT-5.3-Codex |
| Instruction following | 73.5 | 74.6 | Qwen3.6 Max |
| Human preference | – | 64.8 | not comparable |
| Multimodal | 61.2 | – | not comparable |
| Long context | 66.6 | 65.2 | GPT-5.3-Codex |
| Composite | 65.4 | 60.9 | GPT-5.3-Codex |
Where sources disagree
Across the 11 benchmarks both have, GPT-5.3-Codex leads 8, Qwen3.6 Max leads 3. Every overall index that rates both puts GPT-5.3-Codex ahead.
| Overall index | GPT-5.3-Codex (xhigh) | Qwen3.6 Max (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 63.6 | 62.2 | neck and neck |
| AA Intelligence Index v4.3.2 | 32.5 | 28.4 | GPT-5.3-Codex |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.
| Measure | GPT-5.3-Codex (xhigh) | Qwen3.6 Max (max) |
|---|---|---|
| Blended price per 1M tokens | $4.81 | $2.92 |
| Input / output per 1M | $1.75 / $14.00 | $1.30 / $7.80 |
| Cached input per 1M | $0.175 | $0.130 |
| Price per task: chat reply | $0.0049 | $0.0029 |
| Price per task: summarize a 30-page report | $0.029 | $0.020 |
| Price per task: code edit | $0.032 | $0.019 |
| Price per task: agentic coding session | $0.161 | $0.109 |
| Price per task: structured extraction | $0.0063 | $0.0042 |
| Output speed | 103 tok/s | – |
| Time to first answer | 55.9 s | – |
| Time to a complete answer | 60.8 s | – |
Each model is compared at its best-ranked reasoning-effort setting. GPT-5.3-Codex (xhigh) scores 63.6 and takes 56 s to its first answer token. Qwen3.6 Max (max) scores 62.2. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
43 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | GPT-5.3-Codex (xhigh) | Qwen3.6 Max (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 63.6 | 62.2 |
| Agents & tools score | BenchLeader | 80.1 | 72.1 |
| Coding score | BenchLeader | 60.8 | 56.6 |
| Composite score | BenchLeader | 65.4 | 60.9 |
| Instruction following score | BenchLeader | 73.5 | 74.6 |
| Knowledge score | BenchLeader | 66.0 | 62.1 |
| Long context score | BenchLeader | 66.6 | 65.2 |
| Multimodal score | BenchLeader | 61.2 | – |
| Reasoning score | BenchLeader | 68.1 | 58.1 |
| Human preference score | BenchLeader | – | 64.8 |
| Maths score | BenchLeader | – | 64.9 |
| Blended price $/M | BenchLeader | $4.81 | $2.92 |
| Output speed | BenchLeader | 103 tok/s | – |
| Time to first answer | BenchLeader | 55.9 s | – |
| Context window | BenchLeader | 400,000 tokens | 262,144 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | – | 87.4% |
| SimpleBench | SimpleBench | – | 63.0% |
| Arena Hard Prompts | Arena (LMArena) | – | 1483 |
| GPQA Diamond (AA) | Artificial Analysis | 91.5% | 88.8% |
| Humanity's Last Exam (AA) | Artificial Analysis | 42.5% | 30.8% |
| CritPt | Artificial Analysis | 16.9% | 3.7% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | – | 19.0%† |
| LMCA | Epoch AI Benchmarking Hub | – | 42.5% |
| DTBench | Epoch AI Benchmarking Hub | – | 87.2% |
| Coding | |||
| SWE-bench Verified (Epoch) | Epoch AI Benchmarking Hub | 74.8%† | 76.7% |
| WeirdML | WeirdML | 77.9% | – |
| Arena Coding | Arena (LMArena) | – | 1512 |
| Arena WebDev | Arena (LMArena) | 1407† | 1482 |
| LiveCodeBench | Vals AI | 87.3% | – |
| IOI | Vals AI | 53.8% | – |
| SWE-bench (Vals) | Vals AI | 78.0% | 72.8% |
| Vibe Code Bench v1.1 | Vals AI | 61.8% | – |
| ALE-Bench | Epoch AI Benchmarking Hub | 1655.2 | – |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 78.4%† | – |
| Terminal-Bench Hard | Artificial Analysis | 53.0% | 43.9% |
| τ²-Bench Telecom (AA) | Artificial Analysis | 86.0% | 95.9% |
| HiL-Bench | Scale AI SEAL | 4.3%† | – |
| Terminal-Bench 2.0 (Vals) | Vals AI | 64.0%† | 51.7% |
| METR Time Horizons | METR | 74.5%† | – |
| Vending-Bench 2 | Andon Labs | 5940.1† | 4254.2 |
Show all 43 benchmarks
| Metric | Publisher | GPT-5.3-Codex (xhigh) | Qwen3.6 Max (max) |
|---|---|---|---|
| Maths | |||
| OTIS Mock AIME | Epoch AI Benchmarking Hub | – | 91.1% |
| Arena Maths | Arena (LMArena) | – | 1476 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | – | 52.0% |
| AA-Omniscience | Artificial Analysis | 10.9 | 9.2 |
| CorpFin | Vals AI | – | 66.5% |
| CaseLaw v2 | Vals AI | – | 47.9% |
| AA-Omniscience: accuracy | Artificial Analysis | 52.9% | 37.9% |
| AA-Omniscience: non-hallucination | Artificial Analysis | 10.8% | 53.8% |
| Instruction following | |||
| IFBench | Artificial Analysis | 75.4% | 76.6% |
| Arena Instruction Following | Arena (LMArena) | – | 1452 |
| Human preference | |||
| Arena Text | Arena (LMArena) | – | 1461 |
| Arena Creative Writing | Arena (LMArena) | – | 1438 |
| Arena Multi-turn | Arena (LMArena) | – | 1468 |
| Arena Longer Queries | Arena (LMArena) | – | 1474 |
| Multimodal | |||
| MMMU-Pro | Artificial Analysis | 78.5% | – |
| Long context | |||
| AA-LCR | Artificial Analysis | 83.3% | 80.7% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 157.1† | 149.4 |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 32.5 | 28.4 |
Data as of 2026-10-27. Best configuration of each model; every score links to its source on the model pages.
What changed
- Qwen3.6 Max (max): BenchLeader Index 62.8 → 62.3.
- GPT-5.3-Codex (xhigh): BenchLeader Index 64.4 → 63.9.
- GPT-5.3-Codex (xhigh): BenchLeader Index 64.9 → 64.4.
- Qwen3.6 Max (max): BenchLeader Index 63.3 → 62.8.
Related comparisons
- GPT-5.3-Codex vs Claude Opus 5.5
- GPT-5.3-Codex vs Claude Fable 5.1
- GPT-5.3-Codex vs GPT-6 Astra
- Qwen3.6 Max vs Claude Opus 5.5
- Qwen3.6 Max vs Claude Fable 5.1
- Qwen3.6 Max vs GPT-6 Astra
- Qwen3.6 Max vs Muse Wonder Full
- GPT-5.3-Codex vs GPT-6.1 Sol
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
GPT-5.3-Codex vs Qwen3.6 Max: questions
- Is GPT-5.3-Codex better than Qwen3.6 Max?
- GPT-5.3-Codex (xhigh) and Qwen3.6 Max (max) are neck and neck on quality (63.6 vs 62.2, within the ±5.7 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2026-10-27, so check the category scores for your use.
- Which is cheaper, GPT-5.3-Codex or Qwen3.6 Max, and will that change?
- Qwen3.6 Max is cheaper today: $2.92 against $4.81 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is GPT-5.3-Codex better than Qwen3.6 Max for coding?
- GPT-5.3-Codex scores higher in coding (61 vs 57 on the category index, where 50 is average).
- Is GPT-5.3-Codex better than Qwen3.6 Max for agents and tool use?
- GPT-5.3-Codex scores higher in agentic tasks (80 vs 72 on the category index, where 50 is average). On Terminal-Bench Hard, the top shared agents & tools benchmark, it is 53.0% for GPT-5.3-Codex against 43.9% for Qwen3.6 Max.
- Which is better for long documents?
- GPT-5.3-Codex scores higher on long-context benchmarks (67 vs 65). GPT-5.3-Codex accepts more context: 400,000 against 262,144 tokens for Qwen3.6 Max.
- Do independent benchmarks agree on GPT-5.3-Codex vs Qwen3.6 Max?
- Across the 11 benchmarks both have, GPT-5.3-Codex leads 8, Qwen3.6 Max leads 3. Every overall index that rates both puts GPT-5.3-Codex ahead.
- When was Qwen3.6 Max released?
- Qwen3.6 Max was released on 20 Apr 2026.