Find models with documented structured-output support and compare price, context, tools, and provider breadth.
Interactive Tool
| Result | Value | How to read it |
|---|---|---|
| Eligible models | 58 | Models that pass the selected evidence and constraint gates. |
| Ranked shortlist | 20 | Top candidates shown below. |
| Priced candidates | 50 | Models with both input and output price minima. |
| Rank | Model | Fit score | Context | Input / 1M | Workload cost | Published benchmark rank | Providers |
|---|---|---|---|---|---|---|---|
| 1 | Gemini 3.8 FlashGoogle DeepMind | 80.21 | 1.05M tokens | $0.75 / 1M | $150.00 | 7 | 2 |
| 2 | Qwen3.8 MaxQwen | 78.72 | 1M tokens | $1.65 / 1M | $264.02 | 10 | 4 |
| 3 | GLM 5.3 FlashZ.ai | 71.11 | 1M tokens | $0.08 / 1M | $12.50 | 18 | 4 |
| 4 | DeepSeek V4.1 FlashDeepSeek | 68.65 | 1.05M tokens | $0.15 / 1M | $27.00 | 19 | 3 |
| 5 | GPT-5.4 MiniOpenAI | 67.09 | 400K tokens | $0.75 / 1M | $165.00 | 20 | 2 |
| 6 | GPT-5.4 NanoOpenAI | 66.65 | 400K tokens | $0.20 / 1M | $45.00 | 21 | 2 |
| 7 | Gemma 4 31BGoogle DeepMind | 65.78 | 262.14K tokens | $0.20 / 1M | $28.00 | 24 | 5 |
| 8 | Qwen3.7 PlusQwen | 65.07 | 1M tokens | $0.32 / 1M | $57.60 | 23 | 4 |
| 9 | GLM 5.3Z.ai | 62.85 | 1M tokens | $0.90 / 1M | $170.00 | 25 | 4 |
| 10 | Mistral Large 3Mistral AI | 55.84 | 262.14K tokens | $0.25 / 1M | $40.00 | 30 | 2 |
| 11 | Qwen3.7 MaxQwen | 54.61 | 1M tokens | $1.48 / 1M | $236.00 | Unknown | 4 |
| 12 | DeepSeek V4 Flash Vision ExpDeepSeek | 53.87 | 1.05M tokens | $0.22 / 1M | $35.20 | Unknown | 3 |
| 13 | Gemma 4 26B A4BGoogle DeepMind | 53.68 | 262.14K tokens | $0.07 / 1M | $13.00 | 38 | 4 |
| 14 | Qwen3.8 FlashQwen | 53.66 | 1M tokens | $0.11 / 1M | $18.94 | 41 | 4 |
| 15 | Command ACohere | 53.25 | 256K tokens | $2.50 / 1M | $450.00 | 31 | 1 |
| 16 | Gemma 4 E4BGoogle DeepMind | 52.48 | 131.07K tokens | $0.02 / 1M | $4.00 | Unknown | 2 |
| 17 | Qwen3.7 FlashQwen | 52.48 | 1M tokens | $0.03 / 1M | $5.60 | Unknown | 2 |
| 18 | GPT-5 NanoOpenAI | 52.48 | 400K tokens | $0.03 / 1M | $6.50 | Unknown | 2 |
| 19 | GPT-5 MiniOpenAI | 52.38 | 400K tokens | $0.13 / 1M | $32.50 | Unknown | 2 |
| 20 | Qwen3 Coder FlashQwen | 52.35 | 1M tokens | $0.20 / 1M | $39.00 | Unknown | 2 |
Fit scores are transparent screening aids, not universal quality claims. Unknown capability, context, or price fields are never filled by inference.
Inputs
| Input | How it is used |
|---|---|
| Minimum context | Context needed for schema, instructions, examples, and source material. |
| Maximum input price | Optional per-million-token screening ceiling. |
| Tool calling | Whether documented tool support is also required. |
The gate uses the canonical capability fields supplied by primary model or provider records. Ranking favors known context, provider breadth, benchmark coverage, and lower published price.
Documented support says the feature exists; it does not measure JSON validity on your schema. Run a schema-specific evaluation before production selection.
Boundaries
Catalog values retain their source and freshness on the linked model, provider, benchmark, or comparison page. Editable scenario assumptions are not Model Markets measurements.
Continue the Analysis
| Tool | Next question |
|---|---|
| Model Selector | Turn workload constraints into a short, inspectable model shortlist instead of a universal best-model claim. |
| Benchmark Comparison | Inspect like-for-like published results without blending incompatible metrics or versions. |
| Prompt Cost | Make the cost of a system prompt, tool schema, or repeated instruction visible before scale. |
Questions
Shortlist models for schema-constrained JSON and tool-integrated workflows. It returns eligible models ranked by documented support, context fit, provider coverage, benchmark rank, and price.
Where the calculation needs model facts, it uses the current Model Markets catalog snapshot updated 2026-10-19. User-entered assumptions remain clearly editable, and unsupported values stay unknown rather than being inferred.
Capability names can differ by provider endpoint. Strict-schema limits and unsupported keywords are not normalized here. Models with unknown support are excluded rather than assumed incapable. Open the linked canonical records and primary sources before making a production or purchasing decision.
