GPT-5.4 nano vs MiniMax-M3 vs Qwen3.8 Max
Qwen3.8 Max comes out ahead, 83 to 73 and 70 on our weighted score, though GPT-5.4 nano is 6.5× cheaper per token.
Price in rating: Excluded · Include price
OpenAI
GPT-5.4 nano
70/100- ECI146.3
- Price$0.20 / $1.25
- Context400K
MiniMax
MiniMax-M3
73/100- ECI147.9
- Price$0.30 / $1.20
- Context1.05M
- Our pick
Alibaba (Qwen)
Qwen3.8 Max
83/100- ECI157.6
- Price$2.00 / $6.00
- Context1M
Qwen3.8 Max is our pick
Qwen3.8 Max is the better all-round choice, scoring 83/100 against MiniMax-M3 (73) and GPT-5.4 nano (70). It leads on capability and inputs & features. The score weighs capability 67%, inputs & features 20%, context window 13%.
- CapabilityQwen3.8 MaxCapabilities Index (ECI): Qwen3.8 Max 157.6 · MiniMax-M3 147.9 · GPT-5.4 nano 146.3
- Lowest priceGPT-5.4 nanoGPT-5.4 nano $0.463 · MiniMax-M3 $0.525 · Qwen3.8 Max $3.00 per 1M tokens (3:1 blend)
- Longest contextMiniMax-M3MiniMax-M3 1,048,576 · Qwen3.8 Max 1,000,000 · GPT-5.4 nano 400,000 tokens
- Widest inputsQwen3.8 MaxGPT-5.4 nano: Text, Images · MiniMax-M3: Text, Images, Video · Qwen3.8 Max: Text, Images, PDFs, Video
- Self-hostingMiniMax-M3Publishes downloadable weights
| Measure | Weight | GPT-5.4 nano | MiniMax-M3 | Qwen3.8 Max |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 67% | 74 | 75 | 84 |
| Inputs & features | 20% | 70 | 70 | 90 |
| Context window | 13% | 50 | 65 | 64 |
| Overall | 100% | 70/100 | 73/100 | 83/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 146.3 | 147.9 | 157.6 (best) |
| ECI rank | #86 of 158 | #77 of 158 | #30 of 158 (best) |
| GPQA DiamondGraduate-level science questions | 78.5% | 90.9% | 92.7% (best) |
| FrontierMath Tiers 1–3Research-level mathematics | 44.9% | — | 74.7% (best) |
| OTIS Mock AIME 2024–2025Competition mathematics | 87.8% | 71.1% | 99.4% (best) |
| SimpleQA VerifiedShort factual questions | 11.7% | — | 45.8% (best) |
| Price per million tokens | |||
| Input | $0.20 (best) | $0.30 | $2.00 |
| Output | $1.25 | $1.20 (best) | $6.00 |
| Cached input | $0.02 (best) | $0.06 | $0.25 |
| Blended (3:1) | $0.463 (best) | $0.525 | $3.00 |
| Long-context rate | Same rate | Over 512K: $0.60 / $2.40 | Same rate |
| Price source | Official OpenAI API | Official MiniMax (minimax.io) API | Official Alibaba API |
| Limits | |||
| Context window | 400,000 tokens | 1,048,576 tokens (best) | 1,000,000 tokens |
| Max output | 128,000 tokens | 512,000 tokens (best) | 131,072 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | No | No | Yes |
| Audio | No | No | No |
| Video | No | Yes | Yes |
| Reasoning | Yeslow · medium · high · xhigh | Yes | Yeslow · medium · xhigh |
| Tool calling | Yes | Yes | Yes |
| Structured output | Yes | No | Yes |
| Availability | |||
| Weights | Proprietary | Open | Proprietary |
| API model ID | gpt-5.4-nano | MiniMax-M3 | qwen3.8-max |
| API providers | 36 | 52 (best) | 38 |
| Released | Mar 17, 2026 | Jun 1, 2026 | Aug 3, 2026 |
| Knowledge cutoff | Aug 31, 2025 | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
GPT-5.4 nano$4.50
MiniMax-M3$5.40
Qwen3.8 Max$32.00
Which should you choose?
Which is better: GPT-5.4 nano, MiniMax-M3 or Qwen3.8 Max?
Qwen3.8 Max is the better all-round choice, scoring 83/100 against MiniMax-M3 (73) and GPT-5.4 nano (70). It leads on capability and inputs & features. The score weighs capability 67%, inputs & features 20%, context window 13%.
Which is cheaper, GPT-5.4 nano, MiniMax-M3 or Qwen3.8 Max?
GPT-5.4 nano is cheaper at $0.20 input / $1.25 output per million tokens (official OpenAI API price). MiniMax-M3 costs $0.30 input / $1.20 output per million tokens (official MiniMax (minimax.io) API price); Qwen3.8 Max costs $2.00 input / $6.00 output per million tokens (official Alibaba API price). At a typical mix of three input tokens to one output token, that is $0.463 per million tokens for GPT-5.4 nano versus $0.525 for MiniMax-M3 (1.1× as much) and $3.00 for Qwen3.8 Max (6.5× as much).
Which scores higher on benchmarks?
Qwen3.8 Max scores higher on the Capabilities Index (ECI): Qwen3.8 Max 157.6 (#30 of 158), MiniMax-M3 147.9 (#77 of 158) and GPT-5.4 nano 146.3 (#86 of 158). Their confidence ranges do not overlap (155.5–159.9 vs 145.8–150.0), so the gap is a real one. On individual benchmarks: GPQA Diamond — Qwen3.8 Max 92.7%, MiniMax-M3 90.9%, GPT-5.4 nano 78.5%; OTIS Mock AIME 2024–2025 — Qwen3.8 Max 99.4%, GPT-5.4 nano 87.8%, MiniMax-M3 71.1%.
Which is better for coding?
There are no published SWE-bench Verified results for GPT-5.4 nano, MiniMax-M3 and Qwen3.8 Max yet, so there is no like-for-like coding score. On overall capability, Qwen3.8 Max leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.
Which has the bigger context window?
MiniMax-M3 has the largest context window at 1,048,576 tokens, against 1,000,000 for Qwen3.8 Max and 400,000 for GPT-5.4 nano. Maximum output per response: GPT-5.4 nano up to 128,000, MiniMax-M3 up to 512,000, Qwen3.8 Max up to 131,072 tokens.
Which can read images, PDFs, audio or video?
GPT-5.4 nano accepts text and images; MiniMax-M3 accepts text, images and video; Qwen3.8 Max accepts text, images, PDFs and video. Qwen3.8 Max handles the widest range of inputs.
Are any of these open source?
MiniMax-M3 publishes its weights and can be self-hosted; GPT-5.4 nano and Qwen3.8 Max is proprietary.
Which is newer?
Qwen3.8 Max is the newest, released Aug 3, 2026. MiniMax-M3 came out Jun 1, 2026; GPT-5.4 nano came out Mar 17, 2026. Knowledge cutoff: GPT-5.4 nano Aug 31, 2025.
How do you decide the winner?
This rating uses capability (67%, independent benchmark results); inputs & features (20%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (13%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Nov 18, 2026.