Gemini 3.8 Flash vs GPT-5.6 Luna vs Muse Spark 1.3
Too close to call on our weighted score (GPT-5.6 Luna 76, Gemini 3.8 Flash 74, Muse Spark 1.3 73). The right pick depends on what you value most.Too close to call on our weighted score (Gemini 3.8 Flash 86, Muse Spark 1.3 86, GPT-5.6 Luna 81). The right pick depends on what you value most.
Google
Gemini 3.8 Flash
74/100- ECI158.8
- Price$0.75 / $3.75
- Context1.05M
OpenAI
GPT-5.6 Luna
76/100- ECI156.7
- Price$0.20 / $1.20
- Context1.05M
Meta
Muse Spark 1.3
73/100- ECI158.2
- Price$1.25 / $4.25
- Context1.05M
Google
Gemini 3.8 Flash
86/100- ECI158.8
- Price$0.75 / $3.75
- Context1.05M
OpenAI
GPT-5.6 Luna
81/100- ECI156.7
- Price$0.20 / $1.20
- Context1.05M
Meta
Muse Spark 1.3
86/100- ECI158.2
- Price$1.25 / $4.25
- Context1.05M
Too close to call
It is close. Our weighted score puts them within 2 points (GPT-5.6 Luna 76/100, Gemini 3.8 Flash 74/100, Muse Spark 1.3 73/100), so choose by what matters most for your work: Gemini 3.8 Flash for raw capability and GPT-5.6 Luna on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.It is close. Our weighted score puts them within a point (Gemini 3.8 Flash 86/100, Muse Spark 1.3 86/100, GPT-5.6 Luna 81/100), so choose by what matters most for your work: Gemini 3.8 Flash for raw capability and GPT-5.6 Luna on price. The score weighs capability 67%, inputs & features 20%, context window 13%.
- CapabilityGemini 3.8 FlashCapabilities Index (ECI): Gemini 3.8 Flash 158.8 · Muse Spark 1.3 158.2 · GPT-5.6 Luna 156.7
- Lowest priceGPT-5.6 LunaGPT-5.6 Luna $0.45 · Gemini 3.8 Flash $1.50 · Muse Spark 1.3 $2.00 per 1M tokens (3:1 blend)
- Longest contextAbout the sameGPT-5.6 Luna 1,050,000 · Gemini 3.8 Flash 1,048,576 · Muse Spark 1.3 1,048,576 tokens
- Widest inputsGemini 3.8 Flash and Muse Spark 1.3Gemini 3.8 Flash: Text, Images, PDFs, Audio, Video · GPT-5.6 Luna: Text, Images, PDFs · Muse Spark 1.3: Text, Images, PDFs, Audio, Video
- Self-hostingNo confirmed open weightsAll three are available only through APIs
| Measure | Weight | Gemini 3.8 Flash | GPT-5.6 Luna | Muse Spark 1.3 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 86 | 84 | 86 |
| Price | 25% | 40 | 63 | 34 |
| Inputs & features | 15% | 100 | 80 | 100 |
| Context window | 10% | 65 | 65 | 65 |
| Overall | 100% | 74/100 | 76/100 | 73/100 |
| Measure | Weight | Gemini 3.8 Flash | GPT-5.6 Luna | Muse Spark 1.3 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 67% | 86 | 84 | 86 |
| Inputs & features | 20% | 100 | 80 | 100 |
| Context window | 13% | 65 | 65 | 65 |
| Overall | 100% | 86/100 | 81/100 | 86/100 |
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) | 158.8 (best) | 156.7 | 158.2 |
| ECI rank | #25 of 159 (best) | #37 of 159 | #28 of 159 |
| ECI confidence interval | 156.8–161.0 | 154.9–158.8 | 156.2–160.5 |
| ECI evaluation date | Sep 2, 2026 | Jul 9, 2026 | Sep 2, 2026 |
| Documented capabilities | Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows | Cost-efficient GPT-5.6 model for fast, high-volume workloads | Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2. |
| Model parameters | Not published | Not published | Not published |
| Output modalities | Text | Text | Text |
| GPQA DiamondGraduate-level science questions | 95.4%high (best) | 91.6%max | Not published |
| FrontierMath Tiers 1–3Research-level mathematics | 68.4%high | 82.1%max (best) | 74.4%xhigh |
| OTIS Mock AIME 2024–2025Competition mathematics | 98.9%high | 98.3%max | 99.2%xhigh (best) |
| SimpleQA VerifiedShort factual questions | 69.7%high (best) | 41.0%max | Not published |
| Publisher-reported benchmarks | |||
| DeepSWE 1.1high thinking | 73.7pass@1Reported result | Not published | Not published |
| GDPval-AA 2Publisher-reported result | 1545EloReported result | Not published | Not published |
| Finance Agent 2Publisher-reported result | 61.4pass@1Reported result | Not published | Not published |
| Harvey Legal Agent BenchmarkPublisher-reported result | 10all-pass rateReported result | Not published | Not published |
| Terminal-Bench 2.1Publisher-reported result | 89.4pass@1Reported result | Not published | Not published |
| Terminal-Bench 4.0Publisher-reported result | 19.1pass@1Reported result | Not published | Not published |
| GDP.PDFPublisher-reported result | 35all-pass rateReported result | Not published | Not published |
| CharXiv Reasoningwithout tools | 86.2pass@1Reported result | Not published | Not published |
| LVBenchagentic | 87.8pass@1Reported result | Not published | Not published |
| LVBenchstatic; 1024 frames; without tools | 87.1pass@1Reported result | Not published | Not published |
| HLE-Verified · 1811 verified and revised itemsPublisher-reported result | 54.9accuracyReported result | Not published | Not published |
| OSWorld 2.0 · before 2026-08-08 patchbatched tool calls; best of 3 runs; 500 steps | 59partial scoreReported result | Not published | Not published |
| BioMysteryBench · human-solvablewith tools | 88.8pass@1Reported result | Not published | Not published |
| BioMysteryBench · human-difficultwith tools | 56.5pass@1Reported result | Not published | Not published |
| LAB-Bench 2 · 11 subtaskswith tools | 86.2macro-averageReported result | Not published | Not published |
| SWE-Bench ProPublisher-reported result | Not published | 62.7resolve rateReported resultJul 9, 2026 | Not published |
| Terminal-Bench 2.1Publisher-reported result | Not published | 84.7success rateReported resultJul 9, 2026 | Not published |
| DeepSWE 1.1Publisher-reported result | Not published | 67.2resolve rateReported resultJul 9, 2026 | Not published |
| GPQA DiamondPublisher-reported result | Not published | 92.3accuracyReported resultJul 9, 2026 | Not published |
| FrontierMath v2 · Tier 1-3Publisher-reported result | Not published | 78.6accuracyReported resultJul 9, 2026 | Not published |
| BrowseCompPublisher-reported result | Not published | 83.3accuracyReported resultJul 9, 2026 | Not published |
| OSWorld 2.0Publisher-reported result | Not published | 45.6success rateReported resultJul 9, 2026 | Not published |
| MMMU Prono tools | Not published | 78.4accuracyReported resultJul 9, 2026 | Not published |
| Agents' Last ExamPublisher-reported result | Not published | 50.3Reported resultJul 9, 2026 | Not published |
| ToolathlonPublisher-reported result | Not published | 53.4success rateReported resultJul 9, 2026 | Not published |
| Artificial Analysis Intelligence Index 4.1max | Not published | 51.2index scoreReported resultJul 9, 2026 | Not published |
| Artificial Analysis Coding Agent Index 1.1max | Not published | 74.6index scoreReported resultJul 9, 2026 | Not published |
| GDPval-AA 2max effort | Not published | Not published | 1754EloReported resultSep 2, 2026 |
| JobBenchmax effort | Not published | Not published | 64.9mean rubric scoreReported resultSep 2, 2026 |
| OSWorld 2.0 · 08.08max effort | Not published | Not published | 66.9mean partial scoreReported resultSep 2, 2026 |
| OSWorld 2.0 · 08.08max effort | Not published | Not published | 32binary completion rateReported resultSep 2, 2026 |
| DeepSearchQAmax effort | Not published | Not published | 90.3F1Reported resultSep 2, 2026 |
| Agentic IF Index · Meta internalmax effort | Not published | Not published | 57.8scoreReported resultSep 2, 2026 |
| AutomationBench · public v3 task setmax effort | Not published | Not published | 49.6pass@1Reported resultSep 2, 2026 |
| MRCR 2 · 8-needle, 256K-512Kmax effort | Not published | Not published | 98.5mean sequence-match ratioReported resultSep 2, 2026 |
| MRCR 2 · 8-needle, 512K-1Mmax effort | Not published | Not published | 98.1mean sequence-match ratioReported resultSep 2, 2026 |
| DeepSWE 1.1max effort | Not published | Not published | 75.4pass@1Reported resultSep 2, 2026 |
| SWE-Atlas Codebase QnAmax effort | Not published | Not published | 59.4pass@1Reported resultSep 2, 2026 |
| Terminal-Bench 2.1max effort | Not published | Not published | 88.8pass@1Reported resultSep 2, 2026 |
| Price per million tokens | |||
| Input | $0.75 | $0.20 (best) | $1.25 |
| Output | $3.75 | $1.20 (best) | $4.25 |
| Cached input | $0.075 | $0.02 (best) | $0.15 |
| Blended (3:1) | $1.50 | $0.45 (best) | $2.00 |
| Long-context rate | Same rate | Over 272K: $0.40 / $1.80 | Same rate |
| Price source | Official Google API | Official OpenAI API | Official Meta API |
| Limits | |||
| Context window | 1,048,576 tokens | 1,050,000 tokens (best) | 1,048,576 tokens |
| Max input | Not published | 922,000 tokens | Not published |
| Max output | 65,536 tokens | 128,000 tokens | 131,072 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | Yes | Yes | Yes |
| Audio | Yes | No | Yes |
| Video | Yes | No | Yes |
| Reasoning | Yeslow · medium · high | Yeslow · medium · high · xhigh · max | Yesminimal · low · medium · high · xhigh · max |
| Tool calling | Yes | Yes | Yes |
| Structured output | Yes | Yes | Yes |
| File attachments | Yes | Yes | Yes |
| Temperature control | Yes | Yes | Yes |
| Reasoning control | effort | effort | effort |
| Availability | |||
| Weights | Proprietary | Proprietary | Proprietary |
| API model ID | gemini-3.8-flash | gpt-5.6-luna | muse-spark-1.3 |
| API providers | 42 | 57 (best) | 18 |
| Released | Sep 2, 2026 | Jul 9, 2026 | Sep 2, 2026 |
| Knowledge cutoff | Not published | Feb 16, 2026 | Not published |
| Publisher documentation | Publisher documentation | Publisher documentation | Publisher documentation |
| Source metadata updated | Sep 2, 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.
Gemini 3.8 Flash$15.00
GPT-5.6 Luna$4.40
Muse Spark 1.3$21.00
Which should you choose?
Which is better: Gemini 3.8 Flash, GPT-5.6 Luna or Muse Spark 1.3?
It is close. Our weighted score puts them within 2 points (GPT-5.6 Luna 76/100, Gemini 3.8 Flash 74/100, Muse Spark 1.3 73/100), so choose by what matters most for your work: Gemini 3.8 Flash for raw capability and GPT-5.6 Luna on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.It is close. Our weighted score puts them within a point (Gemini 3.8 Flash 86/100, Muse Spark 1.3 86/100, GPT-5.6 Luna 81/100), so choose by what matters most for your work: Gemini 3.8 Flash for raw capability and GPT-5.6 Luna on price. The score weighs capability 67%, inputs & features 20%, context window 13%.
Which is cheaper, Gemini 3.8 Flash, GPT-5.6 Luna or Muse Spark 1.3?
GPT-5.6 Luna is cheaper at $0.20 input / $1.20 output per million tokens (official OpenAI API price). Gemini 3.8 Flash costs $0.75 input / $3.75 output per million tokens (official Google API price); Muse Spark 1.3 costs $1.25 input / $4.25 output per million tokens (official Meta API price). At a typical mix of three input tokens to one output token, that is $0.45 per million tokens for GPT-5.6 Luna versus $1.50 for Gemini 3.8 Flash (3.3× as much) and $2.00 for Muse Spark 1.3 (4.4× as much).
Which scores higher on benchmarks?
Gemini 3.8 Flash scores higher on the Capabilities Index (ECI): Gemini 3.8 Flash 158.8 (#25 of 159), Muse Spark 1.3 158.2 (#28 of 159) and GPT-5.6 Luna 156.7 (#37 of 159). The confidence ranges of the top two overlap (156.8–161.0 vs 156.2–160.5), so treat the gap as small. On individual benchmarks: FrontierMath Tiers 1–3 — GPT-5.6 Luna 82.1%, Muse Spark 1.3 74.4%, Gemini 3.8 Flash 68.4%; OTIS Mock AIME 2024–2025 — Muse Spark 1.3 99.2%, Gemini 3.8 Flash 98.9%, GPT-5.6 Luna 98.3%.
Which is better for coding?
There are no published SWE-bench Verified results for Gemini 3.8 Flash, GPT-5.6 Luna and Muse Spark 1.3 yet, so there is no like-for-like coding score. On overall capability, Gemini 3.8 Flash 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?
Their context windows are effectively the same size: Gemini 3.8 Flash 1,048,576, GPT-5.6 Luna 1,050,000 and Muse Spark 1.3 1,048,576 tokens. Maximum output per response: Gemini 3.8 Flash up to 65,536, GPT-5.6 Luna up to 128,000, Muse Spark 1.3 up to 131,072 tokens.
Which can read images, PDFs, audio or video?
Gemini 3.8 Flash accepts text, images, PDFs, audio and video; GPT-5.6 Luna accepts text, images and PDFs; Muse Spark 1.3 accepts text, images, PDFs, audio and video. Gemini 3.8 Flash handles the widest range of inputs.
Are any of these open source?
No. Gemini 3.8 Flash, GPT-5.6 Luna and Muse Spark 1.3 are proprietary and only available through APIs and apps.
Which is newer?
Gemini 3.8 Flash is the newest, released Sep 2, 2026. Muse Spark 1.3 came out Sep 2, 2026; GPT-5.6 Luna came out Jul 9, 2026. Knowledge cutoff: GPT-5.6 Luna Feb 16, 2026.
How do you decide the winner?
This rating uses 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 Nov 22, 2026.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 22, 2026.