{
    "$schema": "./provider-catalog.schema.json",
    "id": "deepinfra",
    "displayName": "DeepInfra",
    "aliases": [],
    "tier": 2,
    "wire": {
        "baseURL": "https://api.deepinfra.com/v1/openai"
    },
    "models": {
        "default": "deepseek-ai/DeepSeek-V4-Flash-0731",
        "fallbacks": [
            "deepseek-ai/DeepSeek-V4-Pro-0813",
            "openai/gpt-oss-120b",
            "moonshotai/Kimi-K3",
            "zai-org/GLM-5.2",
            "anthropic/claude-haiku-4-5",
            "Qwen/Qwen3-14B",
            "meta-llama/Llama-4-Scout-17B-16E-Instruct"
        ],
        "defaultContextWindow": 128000,
        "defaultMaxOutputTokens": 16384,
        "catalog": {
            "deepseek-ai/DeepSeek-V4-Flash-0731": {
                "contextWindow": 1048576,
                "pricingPerMTok": {
                    "input": 0.06,
                    "output": 0.18,
                    "cachedInput": 0.015
                },
                "vision": false,
                "status": "production",
                "description": "Recommended default — DeepInfra's own tool-calling documentation (docs.deepinfra.com/chat/tool-calling, retrieved 2026-09-28) uses this exact model in its function-calling example (a get_current_weather tool_call round-trip), and its reasoning documentation (docs.deepinfra.com/chat/reasoning) names it among the models that support the reasoning_effort parameter (none/low/medium/high). Live on the unauthenticated GET /v1/openai/models roster (2026-09-28, 187 models) with context_length/max_tokens both 1,048,576 and pricing $0.06/$0.18 per MTok input/output ($0.015 cached-input read). No vision/vlm tag on the live roster. models.defaultMaxOutputTokens (16,384) follows the vendor's documented general cap rather than this per-model figure — see models-level note."
            },
            "deepseek-ai/DeepSeek-V4-Pro-0813": {
                "contextWindow": 1048576,
                "pricingPerMTok": {
                    "input": 1.3,
                    "output": 2.6,
                    "cachedInput": 0.1
                },
                "vision": false,
                "status": "production",
                "description": "DeepSeek's larger Pro variant; roster-listed only (not itself named in a DeepInfra docs example) via the unauthenticated GET /v1/openai/models (2026-09-28) — context 1,048,576, $1.30/$2.60 per MTok input/output, $0.10 cached-input read. Fallback."
            },
            "openai/gpt-oss-120b": {
                "contextWindow": 131072,
                "pricingPerMTok": {
                    "input": 0.037,
                    "output": 0.17
                },
                "vision": false,
                "status": "production",
                "description": "OpenAI's open-weight 120B model hosted on DeepInfra; tagged reasoning + reasoning_effort on the live roster GET (2026-09-28) — context 131,072, $0.037/$0.17 per MTok input/output. Not itself named in DeepInfra's docs examples; fallback."
            },
            "moonshotai/Kimi-K3": {
                "contextWindow": 1048576,
                "pricingPerMTok": {
                    "input": 2.85,
                    "output": 14.25,
                    "cachedInput": 0.285
                },
                "vision": true,
                "status": "production",
                "description": "Named in DeepInfra's tool-calling documentation as \"the current Kimi model in our catalog\" (docs.deepinfra.com/chat/tool-calling). Tagged vlm + vision on the live roster GET (2026-09-28), hence vision: true here (not independently probed for a working vision reply) — context 1,048,576, $2.85/$14.25 per MTok input/output, $0.285 cached-input read. Fallback."
            },
            "zai-org/GLM-5.2": {
                "contextWindow": 1048576,
                "pricingPerMTok": {
                    "input": 0.75,
                    "output": 2.4,
                    "cachedInput": 0.14
                },
                "vision": false,
                "status": "production",
                "description": "Listed by DeepInfra's reasoning documentation (docs.deepinfra.com/chat/reasoning) as one of the models supporting reasoning_effort. Live roster GET (2026-09-28) — context 1,048,576, $0.75/$2.40 per MTok input/output, $0.14 cached-input read. Fallback."
            },
            "anthropic/claude-haiku-4-5": {
                "contextWindow": 200000,
                "pricingPerMTok": {
                    "input": 1.0,
                    "output": 5.0
                },
                "vision": true,
                "status": "production",
                "description": "Anthropic's Claude Haiku 4.5 as served through DeepInfra's model marketplace (owned_by: deepinfra on the live roster). Tagged vlm + vision + reasoning_effort + reasoning on the GET /v1/openai/models roster (2026-09-28), hence vision: true here (not independently probed) — context 200,000, $1.00/$5.00 per MTok input/output. Not itself named in DeepInfra's docs examples; fallback."
            },
            "Qwen/Qwen3-14B": {
                "contextWindow": 40960,
                "pricingPerMTok": {
                    "input": 0.12,
                    "output": 0.24
                },
                "vision": false,
                "status": "production",
                "description": "Smaller Qwen reasoning-tagged model; roster-listed only via the unauthenticated GET /v1/openai/models (2026-09-28) — context 40,960, $0.12/$0.24 per MTok input/output. Fallback."
            },
            "meta-llama/Llama-4-Scout-17B-16E-Instruct": {
                "contextWindow": 327680,
                "pricingPerMTok": {
                    "input": 0.1,
                    "output": 0.3
                },
                "vision": true,
                "status": "production",
                "description": "Meta Llama 4 Scout; tagged vlm + vision on the live roster GET (2026-09-28), hence vision: true here (not independently probed) — context 327,680, $0.10/$0.30 per MTok input/output. Roster-listed only; fallback."
            }
        },
        "topModels": [
            "deepseek-ai/DeepSeek-V4-Flash-0731",
            "deepseek-ai/DeepSeek-V4-Pro-0813",
            "openai/gpt-oss-120b",
            "moonshotai/Kimi-K3",
            "zai-org/GLM-5.2",
            "anthropic/claude-haiku-4-5",
            "Qwen/Qwen3-14B",
            "meta-llama/Llama-4-Scout-17B-16E-Instruct"
        ],
        "visionModel": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
    },
    "capabilities": {
        "text": true,
        "streaming": true,
        "tools": "model-dependent",
        "toolsWithStreaming": true,
        "structuredOutput": true,
        "structuredOutputWithTools": false,
        "embeddings": false,
        "thinking": false
    },
    "errorRules": [],
    "setup": {
        "url": "https://deepinfra.com/dash/api_keys",
        "apiKeyFormat": null,
        "billingPolicy": "no-free-tier",
        "instructions": [
            "1. Visit: https://deepinfra.com/dash/api_keys and sign in (Google, GitHub, email & password, or Corporate SSO — see https://docs.deepinfra.com/account/signing-in)",
            "2. Create an API key and copy it when it is shown",
            "3. No free tier: DeepInfra requires a card or pre-payment before any request succeeds (https://deepinfra.com/pricing)",
            "4. Set {apiKeyEnvVar} in your .env file"
        ],
        "description": "OpenAI-compatible endpoint at https://api.deepinfra.com/v1/openai (docs.deepinfra.com/chat/overview), serving 180+ open-weight and proprietary chat models. No documented auth-failure error body was found on the docs site or the public OpenAPI spec (api.deepinfra.com/openapi.json only documents 200/422 for chat/completions) without a live credentialed probe, so errorRules is empty pending live verification."
    },
    "evidence": {
        "rosterVerified": {
            "date": "2026-09-28",
            "method": "unauthenticated GET https://api.deepinfra.com/v1/openai/models — 187 models; credential-free, no key used",
            "status": 200
        },
        "billingProbe": {
            "date": "2026-09-28",
            "method": "public page https://deepinfra.com/pricing — \"You have to add a card or pre-pay or you won't be able to use our services\"; no free tier documented; credential-free"
        },
        "liveMatrix": null,
        "addedInPR": "https://github.com/juspay/neurolink/pull/1848"
    }
}