GLM-5.2 vs Muse Spark 1.1 vs Qwen3.8 Max 0902
Too close to call on our weighted score (Muse Spark 1.1 66, GLM-5.2 65, Qwen3.8 Max 0902 64). The right pick depends on what you value most.Too close to call on our weighted score (Qwen3.8 Max 0902 80, Muse Spark 1.1 80, GLM-5.2 78). The right pick depends on what you value most.
Z.ai (Zhipu)
GLM-5.2
65/100- ECI153.1
- Price$1.40 / $4.40
- Context1M
Meta
Muse Spark 1.1
66/100- ECI154.9
- Price$1.25 / $4.25
- Context1.05M
Alibaba (Qwen)
Qwen3.8 Max 0902
64/100- ECI156.0
- Price$2.00 / $6.00
- Context1M
Z.ai (Zhipu)
GLM-5.2
78/100- ECI153.1
- Price$1.40 / $4.40
- Context1M
Meta
Muse Spark 1.1
80/100- ECI154.9
- Price$1.25 / $4.25
- Context1.05M
Alibaba (Qwen)
Qwen3.8 Max 0902
80/100- ECI156.0
- Price$2.00 / $6.00
- Context1M
Too close to call
It is close. Our weighted score puts them within 2 points (Muse Spark 1.1 66/100, GLM-5.2 65/100, Qwen3.8 Max 0902 64/100), so choose by what matters most for your work: Qwen3.8 Max 0902 for raw capability, Muse Spark 1.1 on price and Muse Spark 1.1 for long inputs. The score weighs capability 59%, price 29%, context window 12%.It is close. Our weighted score puts them within a point (Qwen3.8 Max 0902 80/100, Muse Spark 1.1 80/100, GLM-5.2 78/100), so choose by what matters most for your work: Qwen3.8 Max 0902 for raw capability, Muse Spark 1.1 on price and Muse Spark 1.1 for long inputs. The score weighs capability 83%, context window 17%.
- CapabilityQwen3.8 Max 0902Capabilities Index (ECI): Qwen3.8 Max 0902 156.0 · Muse Spark 1.1 154.9 · GLM-5.2 153.1
- Lowest priceMuse Spark 1.1Muse Spark 1.1 $2.00 · GLM-5.2 $2.15 · Qwen3.8 Max 0902 $3.00 per 1M tokens (3:1 blend)
- Longest contextMuse Spark 1.1Muse Spark 1.1 1,048,576 · GLM-5.2 1,000,000 · Qwen3.8 Max 0902 1,000,000 tokens
- Widest inputsMuse Spark 1.1 and Qwen3.8 Max 0902GLM-5.2: Text · Muse Spark 1.1: Text, Images, PDFs, Video · Qwen3.8 Max 0902: Text, Images, PDFs, Video
- Self-hostingGLM-5.2Publishes downloadable weights
| Measure | Weight | GLM-5.2 | Muse Spark 1.1 | Qwen3.8 Max 0902 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 59% | 81 | 83 | 84 |
| Price | 29% | 33 | 34 | 26 |
| Context window | 12% | 64 | 65 | 64 |
| Overall | 100% | 65/100 | 66/100 | 64/100 |
Left out because at least one model lacks the data: inputs & features. The remaining weights were rescaled.
| Measure | Weight | GLM-5.2 | Muse Spark 1.1 | Qwen3.8 Max 0902 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 83% | 81 | 83 | 84 |
| Context window | 17% | 64 | 65 | 64 |
| Overall | 100% | 78/100 | 80/100 | 80/100 |
Left out because at least one model lacks the data: inputs & features. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Unavailable specifications are marked “Not published”. Publisher results include their evaluation settings and source.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 153.1 | 154.9 | 156.0 (best) |
| ECI rank | #55 of 159 | #49 of 159 | #42 of 159 (best) |
| ECI confidence interval | 151.3–154.9 | 152.4–157.7 | 153.9–158.3 |
| ECI evaluation date | Jun 16, 2026 | Jul 9, 2026 | Sep 1, 2026 |
| Documented capabilities | Open flagship GLM for long-horizon coding agents and million-token context work | Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration. | 2026-09-02 upgraded snapshot of Qwen3.8 Max with stronger coding, collaborative agents, and multimodal document understanding |
| Model parameters | Not published | Not published | Not published |
| Output modalities | Text | Text | Text |
| GPQA DiamondGraduate-level science questions | 91.9%max | Not published | 92.3%xhigh (best) |
| FrontierMath Tiers 1–3Research-level mathematics | 59.2%max | Not published | 65.6%xhigh (best) |
| OTIS Mock AIME 2024–2025Competition mathematics | 86.4%max | Not published | 100%xhigh (best) |
| SWE-bench VerifiedFixing real GitHub issues | 78.7%max | Not published | Not published |
| SimpleQA VerifiedShort factual questions | 34.2%max | 57.8% (best) | 47.3%xhigh |
| Publisher-reported benchmarks | |||
| SWE-Bench ProPublisher-reported result | 62.1resolve rateReported resultJun 16, 2026 | 61.5resolve rateReported resultJul 9, 2026 | Not published |
| Terminal-Bench 2.1Publisher-reported result | 82.7success rateReported resultJun 16, 2026 | 80success rateReported resultJul 9, 2026 | Not published |
| FrontierSWEPublisher-reported result | 74.4dominanceReported resultJun 16, 2026 | Not published | Not published |
| Humanity's Last Exam · text-only subsetPublisher-reported result | 40.5accuracyReported resultJun 16, 2026 | Not published | Not published |
| Humanity's Last Exam · text-only subsetwith tools | 54.7accuracyReported resultJun 16, 2026 | Not published | Not published |
| CritPtPublisher-reported result | 20.9accuracyReported resultJun 16, 2026 | Not published | Not published |
| AIME 2026Publisher-reported result | 99.2accuracyReported resultJun 16, 2026 | Not published | Not published |
| HMMT November 2025Publisher-reported result | 94.4accuracyReported resultJun 16, 2026 | Not published | Not published |
| HMMT February 2026Publisher-reported result | 92.5accuracyReported resultJun 16, 2026 | Not published | Not published |
| IMOAnswerBenchPublisher-reported result | 91accuracyReported resultJun 16, 2026 | Not published | Not published |
| GPQA DiamondPublisher-reported result | 91.2accuracyReported resultJun 16, 2026 | Not published | Not published |
| NL2RepoPublisher-reported result | 48.9resolve rateReported resultJun 16, 2026 | Not published | Not published |
| DeepSWEPublisher-reported result | 46.2resolve rateReported resultJun 16, 2026 | Not published | Not published |
| Program BenchPublisher-reported result | 63.7scoreReported resultJun 16, 2026 | Not published | Not published |
| PostTrainBenchPublisher-reported result | 34.3scoreReported resultJun 16, 2026 | Not published | Not published |
| SWE MarathonPublisher-reported result | 13resolve rateReported resultJun 16, 2026 | Not published | Not published |
| MCP Atlas · public subsetPublisher-reported result | 76.8scoreReported resultJun 16, 2026 | Not published | Not published |
| Tool-DecathlonPublisher-reported result | 48.2scoreReported resultJun 16, 2026 | Not published | Not published |
| DeepSWE 1.1Publisher-reported result | Not published | 53.3resolve rateReported resultJul 9, 2026 | Not published |
| MCP AtlasPublisher-reported result | Not published | 88.1success rateReported resultJul 9, 2026 | Not published |
| JobBenchPublisher-reported result | Not published | 54.7success rateReported resultJul 9, 2026 | Not published |
| Toolathlon-VerifiedPublisher-reported result | Not published | 75.6success rateReported resultJul 9, 2026 | Not published |
| Humanity's Last Examwith tools | Not published | 62.1accuracyReported resultJul 9, 2026 | Not published |
| OSWorld-VerifiedPublisher-reported result | Not published | 80.8success rateReported resultJul 9, 2026 | Not published |
| Finance Agent v2Publisher-reported result | Not published | 57.2accuracyReported resultJul 9, 2026 | Not published |
| CharXiv ReasoningPublisher-reported result | Not published | 88.4accuracyReported resultJul 9, 2026 | Not published |
| BabyVisionPublisher-reported result | Not published | 76.3accuracyReported resultJul 9, 2026 | Not published |
| Price per million tokens | |||
| Input | $1.40 | $1.25 (best) | $2.00 |
| Output | $4.40 | $4.25 (best) | $6.00 |
| Cached input | $0.26 | $0.15 (best) | Not published |
| Blended (3:1) | $2.15 | $2.00 (best) | $3.00 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Z.AI API | Official Meta API | Median of 9 providers |
| Limits | |||
| Context window | 1,000,000 tokens | 1,048,576 tokens (best) | 1,000,000 tokens |
| Max input | Not published | Not published | Not published |
| Max output | 131,072 tokens | 131,072 tokens | 131,072 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | Yes |
| PDFs | No | Yes | Yes |
| Audio | No | No | No |
| Video | No | Yes | Yes |
| Reasoning | Yeshigh · max | Yesminimal · low · medium · high · xhigh | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | Yes | Yes | Not published |
| File attachments | No | Yes | Yes |
| Temperature control | Yes | Yes | Yes |
| Reasoning control | effort | effort | always |
| Availability | |||
| Weights | Open | Proprietary | Proprietary |
| API model ID | glm-5.2 | muse-spark-1.1 | Not published |
| API providers | 88 (best) | 15 | 9 |
| Released | Jun 13, 2026 | Jul 9, 2026 | Sep 2, 2026 |
| Knowledge cutoff | Not published | Not published | Not published |
| Publisher documentation | Publisher documentation · Model weights | Publisher documentation | Not published |
| Source metadata updated | Jun 13, 2026 | Jul 9, 2026 | Sep 2, 2026 |
| Data sources | models.dev · Epoch AI | models.dev · Epoch AI | models.dev · Epoch AI |
| Catalog checked | Nov 22, 2026 | Nov 22, 2026 | Nov 22, 2026 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
GLM-5.2$22.80
Muse Spark 1.1$21.00
Qwen3.8 Max 0902$32.00
Which should you choose?
Which is better: GLM-5.2, Muse Spark 1.1 or Qwen3.8 Max 0902?
It is close. Our weighted score puts them within 2 points (Muse Spark 1.1 66/100, GLM-5.2 65/100, Qwen3.8 Max 0902 64/100), so choose by what matters most for your work: Qwen3.8 Max 0902 for raw capability, Muse Spark 1.1 on price and Muse Spark 1.1 for long inputs. The score weighs capability 59%, price 29%, context window 12%.It is close. Our weighted score puts them within a point (Qwen3.8 Max 0902 80/100, Muse Spark 1.1 80/100, GLM-5.2 78/100), so choose by what matters most for your work: Qwen3.8 Max 0902 for raw capability, Muse Spark 1.1 on price and Muse Spark 1.1 for long inputs. The score weighs capability 83%, context window 17%.
Which is cheaper, GLM-5.2, Muse Spark 1.1 or Qwen3.8 Max 0902?
Muse Spark 1.1 is cheaper at $1.25 input / $4.25 output per million tokens (official Meta API price). GLM-5.2 costs $1.40 input / $4.40 output per million tokens (official Z.AI API price); Qwen3.8 Max 0902 costs $2.00 input / $6.00 output per million tokens (median across 9 API providers). At a typical mix of three input tokens to one output token, that is $2.00 per million tokens for Muse Spark 1.1 versus $2.15 for GLM-5.2 (1.1× as much) and $3.00 for Qwen3.8 Max 0902 (1.5× as much).
Which scores higher on benchmarks?
Qwen3.8 Max 0902 scores higher on the Capabilities Index (ECI): Qwen3.8 Max 0902 156.0 (#42 of 159), Muse Spark 1.1 154.9 (#49 of 159) and GLM-5.2 153.1 (#55 of 159). The confidence ranges of the top two overlap (153.9–158.3 vs 152.4–157.7), so treat the gap as small. On individual benchmarks: SimpleQA Verified — Muse Spark 1.1 57.8%, Qwen3.8 Max 0902 47.3%, GLM-5.2 34.2%.
Which is better for coding?
There are no published SWE-bench Verified results for Muse Spark 1.1 and Qwen3.8 Max 0902 yet, so there is no like-for-like coding score. On overall capability, Qwen3.8 Max 0902 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?
Muse Spark 1.1 has the largest context window at 1,048,576 tokens, against 1,000,000 for GLM-5.2 and 1,000,000 for Qwen3.8 Max 0902. Maximum output per response: GLM-5.2 up to 131,072, Muse Spark 1.1 up to 131,072, Qwen3.8 Max 0902 up to 131,072 tokens.
Which can read images, PDFs, audio or video?
GLM-5.2 accepts text; Muse Spark 1.1 accepts text, images, PDFs and video; Qwen3.8 Max 0902 accepts text, images, PDFs and video. Muse Spark 1.1 handles the widest range of inputs.
Are any of these open source?
GLM-5.2 publishes its weights and can be self-hosted; Muse Spark 1.1 and Qwen3.8 Max 0902 is proprietary.
Which is newer?
Qwen3.8 Max 0902 is the newest, released Sep 2, 2026. Muse Spark 1.1 came out Jul 9, 2026; GLM-5.2 came out Jun 13, 2026.
How do you decide the winner?
This rating uses capability (59%, independent benchmark results); price (29%, blended price per million tokens (3 input : 1 output), log scale); context window (12%, 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 22, 2026.This rating uses capability (83%, independent benchmark results); context window (17%, 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 22, 2026.