Grok 4.6 vs MiniCPM5-2B vs Qwen3.8 Max
Too close to call on our weighted score (MiniCPM5-2B 53, Qwen3.8 Max 53, Grok 4.6 44). The right pick depends on what you value most.
xAIGrok 4.6
44/100- ECI156.6
- Price$2.00 / $6.00
- Context500K
OpenBMB
MiniCPM5-2B
53/100- ECI—
- Price$0.124 / $0.743
- Context131K
Alibaba (Qwen)Qwen3.8 Max
53/100- ECI156.6
- Price$2.00 / $6.00
- Context1M
Too close to call
It is close. Our weighted score puts them within a point (MiniCPM5-2B 53/100, Qwen3.8 Max 53/100, Grok 4.6 44/100), so choose by what matters most for your work: MiniCPM5-2B on price and Qwen3.8 Max for long inputs. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceMiniCPM5-2BMiniCPM5-2B $0.279 · Grok 4.6 $3.00 · Qwen3.8 Max $3.00 per 1M tokens (3:1 blend)
- Longest contextQwen3.8 MaxQwen3.8 Max 1,000,000 · Grok 4.6 500,000 · MiniCPM5-2B 131,072 tokens
- Widest inputsQwen3.8 MaxGrok 4.6: Text, Images · MiniCPM5-2B: Text · Qwen3.8 Max: Text, Images, PDFs, Video
- Self-hostingMiniCPM5-2BPublishes downloadable weights (apache-2.0)
| Measure | Weight | Grok 4.6 | MiniCPM5-2B | Qwen3.8 Max |
|---|---|---|---|---|
| Price | 50% | 27 | 76 | 27 |
| Inputs & features | 30% | 70 | 35 | 90 |
| Context window | 20% | 48 | 24 | 60 |
| Overall | 100% | 44/100 | 53/100 | 53/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Grok 4.6xAI | MiniCPM5-2BOpenBMB | Qwen3.8 MaxAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 156.6 (best) | — | 156.6 |
| ECI rank | #19 of 148 (best) | — | #20 of 148 |
| GPQA DiamondGraduate-level science questions | 94.0% (best) | — | 92.7% |
| FrontierMath Tiers 1–3Research-level mathematics | 66.0% | — | 74.7% (best) |
| OTIS Mock AIME 2024–2025Competition mathematics | 99.2% | — | 99.4% (best) |
| SimpleQA VerifiedShort factual questions | 49.3% (best) | — | 45.8% |
| Price per million tokens | |||
| Input | $2.00 | $0.124 (best) | $2.00 |
| Output | $6.00 | $0.743 (best) | $6.00 |
| Cached input | $0.50 | — | $0.25 (best) |
| Blended (3:1) | $3.00 | $0.279 (best) | $3.00 |
| Long-context rate | Over 200K: $4.00 / $12.00 | Same rate | Same rate |
| Price source | Official xAI API | Median of 1 providers | Official Alibaba API |
| Limits | |||
| Context window | 500,000 tokens | 131,072 tokens | 1,000,000 tokens (best) |
| Max output | 500,000 tokens (best) | 131,072 tokens | 131,072 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | No | Yes |
| PDFs | No | No | Yes |
| Audio | No | No | No |
| Video | No | No | Yes |
| Reasoning | Yeslow · medium · high · xhigh | Yes | Yeslow · medium · xhigh |
| Tool calling | Yes | Yes | Yes |
| Structured output | Yes | No | Yes |
| Availability | |||
| Weights | Proprietary | Openapache-2.0 | Proprietary |
| API model ID | grok-4.6 | — | qwen3.8-max |
| API providers | 29 (best) | 1 | 25 |
| Released | Aug 12, 2026 | Sep 6, 2026 | Aug 3, 2026 |
| Knowledge cutoff | Feb 1, 2026 | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Grok 4.6$32.00
- MiniCPM5-2B$2.73
- Qwen3.8 Max$32.00
Which should you choose?
Which is better: Grok 4.6, MiniCPM5-2B or Qwen3.8 Max?
It is close. Our weighted score puts them within a point (MiniCPM5-2B 53/100, Qwen3.8 Max 53/100, Grok 4.6 44/100), so choose by what matters most for your work: MiniCPM5-2B on price and Qwen3.8 Max for long inputs. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Grok 4.6, MiniCPM5-2B or Qwen3.8 Max?
MiniCPM5-2B is cheaper at $0.124 input / $0.743 output per million tokens (median across 1 API provider). Grok 4.6 costs $2.00 input / $6.00 output per million tokens (official xAI 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 $0.279 per million tokens for MiniCPM5-2B versus $3.00 for Grok 4.6 (11× as much) and $3.00 for Qwen3.8 Max (11× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Grok 4.6 has an ECI of 156.6, MiniCPM5-2B has not been scored yet and Qwen3.8 Max has an ECI of 156.6.
Which is better for coding?
There are no published SWE-bench Verified results for Grok 4.6, MiniCPM5-2B and Qwen3.8 Max yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.
Which has the bigger context window?
Qwen3.8 Max has the largest context window at 1,000,000 tokens, against 500,000 for Grok 4.6 and 131,072 for MiniCPM5-2B. Maximum output per response: Grok 4.6 up to 500,000, MiniCPM5-2B up to 131,072, Qwen3.8 Max up to 131,072 tokens.
Which can read images, PDFs, audio or video?
Grok 4.6 accepts text and images; MiniCPM5-2B accepts text; Qwen3.8 Max accepts text, images, PDFs and video. Qwen3.8 Max handles the widest range of inputs.
Are any of these open source?
MiniCPM5-2B publishes its weights (apache-2.0) and can be self-hosted; Grok 4.6 and Qwen3.8 Max is proprietary.
Which is newer?
MiniCPM5-2B is the newest, released Sep 6, 2026. Grok 4.6 came out Aug 12, 2026; Qwen3.8 Max came out Aug 3, 2026. Knowledge cutoff: Grok 4.6 Feb 1, 2026.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, 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 Oct 4, 2026.
