Qwen3.5 397B A17B vs GPT-5.4
At a Glance
| Compare |  GPT-5.4OpenAI | |
|---|---|---|
| Intelligence, Cost, and Efficiency | ||
| IntelligenceHigher is better · MM Intelligence v2.5 | UnrankedNot in the 46-model eligible cohort | #14 of 4672.8 score · 3/3 sources · complete |
| CostLower is better · Published-token output estimate | UnrankedNot in the 44-model eligible cohort | #36 of 44$0.274 per LiveBench case |
| EfficiencyHigher is better · MM Efficiency v1.5 | UnrankedNot in the 38-model eligible cohort | #25 of 3850.1 score · 3/3 sources · complete |
| Pricing and Limits | ||
| Input priceFrom · USD / 1M tokens | $0.45Deepinfra ↗ · Sep 21, 2026 | $2.50Openai ↗ · Sep 3, 2026 |
| Output priceFrom · USD / 1M tokens | $3.00Deepinfra ↗ · Sep 21, 2026 | $15.00Openai ↗ · Sep 3, 2026 |
| Context windowMaximum documented tokens | 262K | 1,050K |
| Model facts checked | Aug 28, 2026View model evidence → | Sep 3, 2026View model evidence → |
Token prices are the lowest available sourced USD rates; input and output may use different providers. Cost ranking estimates output spend on LiveBench, not a full request bill. Ranking methodology →
Available Benchmarks
| Benchmark | Qwen3.5-397B-A17B | GPT-5.4 |
|---|---|---|
| LMArena Text Arenatext-2026-09-13-d25aabda0010 · arena_rating · leader | 1,438.2699% of row best · rating · qwen3.5-397b-a17b; 95% CI [1434.97391785, 1441.55191176]; votes 77007; rank 71 | 1,452.64100% of row best · rating · gpt-5.4; 95% CI [1448.87114809, 1456.41035855]; votes 63526; rank 41 |
| LMArena Vision Arenavision-2026-09-13-d25aabda0010 · arena_rating · leader | 1,265.1598% of row best · rating · qwen3.5-397b-a17b; 95% CI [1259.44632362, 1270.86149272]; votes 27569; rank 42 | 1,292.51100% of row best · rating · gpt-5.4; 95% CI [1285.77825721, 1299.23452617]; votes 21188; rank 22 |
| Overall ResultCounted from the protocol-matched rows above | 0 benchmark winsNo overall winner | 2 benchmark winsNo overall winner |
Third-party benchmark Only like-for-like primary-publisher results are shown; raw scores, relative scores, configuration, and token spend remain visible.
Side-by-Side Facts
| Field | Qwen3.5-397B-A17B | GPT-5.4 |
|---|---|---|
| Developer | Qwen | OpenAI |
| Family | Qwen3 5 397b A17b | Gpt 5 4 |
| Model | Qwen3.5-397B-A17B | GPT-5.4 |
| Version | Qwen3.5-397B-A17B | GPT-5.4 |
| Lifecycle | active | active |
| Released | 2026-02-15 | 2026-03-05 |
| Knowledge cutoff | Unknown | 2025-08-31 |
| Input modalities | Text, Image | Text, Image |
| Output modalities | Text | Text |
| Context window | 262K | 1,050K |
| Total parameters | 403.4B | Unknown |
| Active parameters | 17B | Unknown |
| License | apache-2.0 | Unknown |
| Open weights | Yes | No |
| API available | Yes | Yes |
| Self-hostable | Yes | No |
| Provider access | Deepinfra (Standard), Hugging Face (Standard), Openrouter (Standard), Together Ai (Standard) | Openai (Standard), Openrouter (Standard) |
| Capabilities | chat, generation, reasoning, tools | chat, generation, reasoning, structured_outputs, tools |
Qwen3.5 397B A17B Capabilities
GPT-5.4 Capabilities
Related Comparisons
All image comparisons →| A | Pair | B | Context |
|---|---|---|---|
| vs | GPT-6 AstraOpenAI | cross-developer peersimage, text | |
| vs | Claude Fable 5.1Anthropic | cross-developer peersimage, text | |
| vs |  Gemini 3.1 ProGoogle DeepMind | cross-developer peerstext | |
| vs |  DeepSeek-V4.1-FlashDeepSeek | cross-developer peersimage, text | |
| vs | Grok 4.6xAI | cross-developer peersimage, text | |
| vs |  Qwen3.8-MaxQwen | family variantsimage, text | |
| vs |  Kimi-K3Moonshot AI | cross-developer peersimage, text | |
 MiniMax-M3MiniMax | vs | cross-developer peersimage, text | |
| vs |  GLM-5.3Z.ai | cross-developer peerstext | |
| vs |  Hy4 previewTencent | cross-developer peerstext | |
| vs | Seed 2.1 ProByteDance Seed | cross-developer peersimage, text | |
| vs |  Mistral Large 3Mistral AI | cross-developer peersimage, text |
Primary Evidence
Sources and Freshness
Questions
Qwen3.5 397B A17B vs GPT-5.4 FAQs
Is Qwen3.5 397B A17B or GPT-5.4 better for coding?+
This comparison does not currently contain a protocol-matched coding benchmark for both Qwen3.5 397B A17B and GPT-5.4, so Model Markets cannot name a coding leader from pricing, context size, or capability labels alone.
Which is cheaper, Qwen3.5 397B A17B or GPT-5.4?+
Qwen3.5 397B A17B is $0.45 and GPT-5.4 is $2.50 per million tokens, so Qwen3.5 397B A17B is cheaper on this metric. Qwen3.5 397B A17B is $3.00 and GPT-5.4 is $15.00 per million tokens, so Qwen3.5 397B A17B is cheaper on this metric.
Which has a larger context window, Qwen3.5 397B A17B or GPT-5.4?+
GPT-5.4 has the larger sourced context window. Qwen3.5 397B A17B supports 262K and GPT-5.4 supports 1,050K.
Which performs better in benchmarks, Qwen3.5 397B A17B or GPT-5.4?+
There is no overall benchmark winner: An overall winner requires at least two decisive benchmarks from at least two original publishers.
Can Qwen3.5 397B A17B or GPT-5.4 be self-hosted?+
Qwen3.5 397B A17B is the only model in this pair currently marked as self-hostable. Qwen3.5 397B A17B is open weight; GPT-5.4 is not marked open weight.
Can Qwen3.5 397B A17B and GPT-5.4 understand images?+
Qwen3.5 397B A17B is documented with image input; GPT-5.4 is documented with image input. This reflects supported input modalities, not vision quality.
Which can generate longer answers, Qwen3.5 397B A17B or GPT-5.4?+
Neither has a larger sourced maximum output. Qwen3.5 397B A17B is — and GPT-5.4 is 128K.
Do Qwen3.5 397B A17B and GPT-5.4 support reasoning and tool use?+
Qwen3.5 397B A17B: reasoning, tool calling, and image input. GPT-5.4: reasoning, tool calling, and image input. Feature support does not establish relative quality.
Which is available from more inference providers, Qwen3.5 397B A17B or GPT-5.4?+
Qwen3.5 397B A17B has 4 sourced provider routes; GPT-5.4 has 2, so Qwen3.5 397B A17B has broader tracked availability.
Which offers better value, Qwen3.5 397B A17B or GPT-5.4?+
There is no universal value winner. Compare the input and output prices above with the matched benchmark result for your workload: cheaper tokens can be offset by different quality, token usage, latency, or provider availability.