Deep Infra
Ultra-cheap open-weight inference — Llama 4, DeepSeek V4 Pro, Qwen3 at rock-bottom prices
Deep Infra is an inference-focused API provider that consistently offers the lowest per-token prices for open-weight models. The 2026 catalog includes Llama 4 Scout/Maverick, DeepSeek V4 Pro, Qwen3-235B, Kimi K3, and the full Llama 3.x family. Deep Infra is the go-to choice for cost-sensitive production inference where price is the primary constraint.
The most price-competitive inference API for open-weight models — Llama 4, DeepSeek V4 Pro, and Qwen3 at prices 30–50% below comparable providers, with zero migration friction via OpenAI-compatible API.
- Consistently lowest prices for open-weight models — often 30–50% below competitors
- 2026 catalog: Llama 4, DeepSeek V4 Pro, Qwen3-235B, Kimi K3
- Wide model catalog updated rapidly as new models release
- OpenAI-compatible API — zero migration friction
- No rate limits on most models
- Fast cold start times
- No proprietary models — open-weight only
- No enterprise SLAs or dedicated support
- Smaller ecosystem than Together AI or Fireworks
Deep Infra Models
| Model | Modality | Tier | Price | Cached /1M | TTFT | Throughput | Context | Intel. | Caps | |
|---|---|---|---|---|---|---|---|---|---|---|
Meta-Llama-3.1-8B-Instruct Llama 3.1 | Chat | budget | $0.02/ $0.05 in / out per 1M EST Jul 2026 | — | 60ms | 1.1K t/s | 131K | 38 | ||
Qwen3-32B Qwen3 | Chat | efficient | $0.08/ $0.28 in / out per 1M EST Oct 2026 | — | — | — | 41K | 70 | ||
Qwen3-235B-A22B-Instruct-2507 Qwen3 | Chat | advanced | $0.09/ $0.55 in / out per 1M EST Oct 2026 | — | — | — | 262K | 83 | ||
Qwen3-Next-80B-A3B-Instruct Qwen3 Next | Chat | efficient | $0.09/ $1.10 in / out per 1M EST Oct 2026 | — | — | — | 262K | 74 | ||
Llama-3.3-70B-Instruct Llama 3.3 | Chat | efficient | $0.10/ $0.28 in / out per 1M EST Jul 2026 | — | 280ms | 280 t/s | 131K | 64 | ||
Qwen3.5-9B Qwen3.5 | Chat | budget | $0.10/ $0.15 in / out per 1M EST Oct 2026 | — | — | — | 262K | 68 | ||
Qwen3.6-35B-A3B Qwen3.6 | Chat | efficient | $0.10/ $0.95 in / out per 1M EST Oct 2026 | $0.10 | — | — | 262K | 80 | ||
Qwen3-30B-A3B Qwen3 | Chat | efficient | $0.12/ $0.50 in / out per 1M EST Oct 2026 | — | — | — | 41K | 68 | ||
Qwen3-14B Qwen3 | Chat | budget | $0.12/ $0.24 in / out per 1M EST Oct 2026 | — | — | — | 41K | 62 | ||
DeepSeek-V3 DeepSeek V3 | Chat | advanced | $0.14/ $0.28 in / out per 1M EST Jul 2026 | — | 500ms | 120 t/s | 164K | 80 | ||
Qwen3.5-35B-A3B Qwen3.5 | Chat | efficient | $0.14/ $1.00 in / out per 1M EST Oct 2026 | $0.05 | — | — | 262K | 78 | ||
Qwen3-VL-30B-A3B-Instruct Qwen3 VL | Chat | efficient | $0.15/ $0.60 in / out per 1M EST Oct 2026 | — | — | — | 262K | 72 | ||
Qwen3-VL-235B-A22B-Instruct Qwen3 VL | Chat | advanced | $0.20/ $0.88 in / out per 1M EST Oct 2026 | $0.11 | — | — | 262K | 82 | ||
DeepSeek-R1-Distill-Llama-70B DeepSeek R1 | Chat | efficient | $0.23/ $0.69 in / out per 1M EST Jul 2026 | — | 700ms | 100 t/s | 66K | 72 | ||
Mixtral-8x7B-Instruct-v0.1 Mixtral | Chat | budget | $0.24/ $0.24 in / out per 1M EST Jul 2026 | — | 200ms | 350 t/s | 33K | 52 | ||
Qwen3.5-27B Qwen3.5 | Chat | efficient | $0.26/ $2.60 in / out per 1M EST Oct 2026 | — | — | — | 262K | 76 | ||
Qwen3-Coder-480B-A35B-Instruct-Turbo Qwen3 Coder | Chat | advanced | $0.30/ $1.00 in / out per 1M EST Oct 2026 | $0.10 | — | — | 262K | 86 | ||
Qwen3.6-27B Qwen3.6 | Chat | efficient | $0.32/ $3.20 in / out per 1M EST Oct 2026 | — | — | — | 262K | 78 | ||
Qwen2.5-72B-Instruct Qwen2.5 | Chat | efficient | $0.36/ $0.40 in / out per 1M EST Oct 2026 | — | — | — | 33K | 66 | ||
Qwen3.5-397B-A17B Qwen3.5 | Chat | advanced | $0.45/ $3.00 in / out per 1M EST Oct 2026 | $0.22 | — | — | 262K | 88 | ||
DeepSeek-R1 DeepSeek R1 | Chat | advanced | $0.55/ $2.19 in / out per 1M EST Jul 2026 | — | 900ms | 80 t/s | 164K | 85 | ||
Qwen3.8-Max Qwen3.8 | Chat | frontier | $1.65/ $4.95 in / out per 1M EST Oct 2026 | $0.21 | — | — | 250K | 88 | ||
Qwen3.7-Max Qwen3.7 | Chat | frontier | $2.50/ $7.50 in / out per 1M EST Oct 2026 | $0.50 | — | — | 250K | 86 |
Deep Infra — Frequently Asked Questions (FAQs)
Similar Providers
Deep Infra Reviews
All LLM Providers
Full comparison tableDeep Infra LLM pricing overview
Deep Infra is an inference API provider that hosts open-weight and third-party large language models. Deep Infra is an inference-focused API provider that consistently offers the lowest per-token prices for open-weight models. The 2026 catalog includes Llama 4 Scout/Maverick, DeepSeek V4 Pro, Qwen3-235B, Kimi K3, and the full Llama 3.x family. Deep Infra is the go-to choice for cost-sensitive production inference where price is the primary constraint. Common use cases include Cost-sensitive inference, High-volume batch processing, Llama 4 production, DeepSeek hosting. Headquartered in San Francisco, CA, founded 2023. All models are accessible via a REST API compatible with standard OpenAI-style request formats, enabling drop-in integration with most LLM frameworks and orchestration tools.
Deep Infra vs other LLM providers
Deep Infra competes with OpenAI, Anthropic, Google, Groq, Together AI, Mistral, Cohere, and other inference API providers across dimensions of price, throughput, latency, context window, and model intelligence. The most price-competitive inference API for open-weight models — Llama 4, DeepSeek V4 Pro, and Qwen3 at prices 30–50% below comparable providers, with zero migration friction via OpenAI-compatible API. Use the LLM provider comparison tool to see Deep Infra token pricing, latency, and throughput side-by-side with any other provider. The full LLM pricing table shows all providers ranked by input token cost, output token cost, and throughput in a single sortable view.
Why choose Deep Infra?
Deep Infra stands out because: The most price-competitive inference API for open-weight models — Llama 4, DeepSeek V4 Pro, and Qwen3 at prices 30–50% below comparable providers, with zero migration friction via OpenAI-compatible API.. The API currently offers 23 models, spanning 2 frontier, 6 advanced, 11 efficient, 4 budget tiers. For teams running high-volume inference workloads, prompt caching and batch API endpoints can reduce effective input token costs by 50–90% — check the context window cost guide for a full breakdown of caching economics.
Understanding Deep Infra token pricing
Deep Infra charges separately for input (prompt) and output (completion) tokens, priced per 1M tokens in USD. Output tokens are typically 3–5× more expensive than input tokens due to the compute cost of autoregressive generation. The cheapest {meta.name} input rate is currently $0.02/1M tokens. Prices shown are sourced from Deep Infra's public pricing page and updated every 15 minutes. Need help estimating your spend? Read the LLM API cost calculator guide — it covers tokens, context windows, prompt caching, and batch discounts with worked examples for RAG, chat history, and document processing workloads.
Deep Infra context window and model capabilities
Context window size directly affects both capability and cost — every token in the context window is charged as an input token. Deep Infra's largest context window is 262K tokens. For RAG and document processing workloads, longer context windows enable richer retrieval but increase per-call costs proportionally. Prompt caching — where supported — stores the KV state of repeated prefixes and charges 75–90% less for cache hits, making it the most impactful cost optimization for applications with consistent system prompts or retrieved documents. See the LLM context window cost guide for a full analysis of how context length affects your API bill.
Self-hosted vs managed inference: when Deep Infra makes sense
Managed inference APIs like Deep Infra eliminate infrastructure overhead — no GPU provisioning, driver management, or model serving stack to maintain. The trade-off is cost at scale: a single H100 at ~$2.50/hr can serve ~500K tokens/min of Llama 3.3 70B, which at Deep Infra API rates would cost significantly more per token. The break-even point depends on your request volume, latency requirements, and engineering capacity. For teams processing fewer than ~10M tokens/day, managed APIs are almost always cheaper when total cost of ownership is considered. Above that threshold, self-hosted inference on rented GPU compute typically wins on unit economics. Read the cheapest GPU cloud guide for a full break-even analysis. Historical Deep Infra token price data is available in the LLM price history charts.
Compare Deep Infra With Another LLM Provider
Compare Deep Infra vs other providers
Side-by-side token pricing, latency, throughput, and context window. View all LLM provider comparisons →
Looking for alternatives?
Compare Deep Infra with other LLM API providers side by side.