Every model, one key.

Flagship models servable through RawLine — each research-ready (jailbreak-style presets for red-team eval) behind a single OpenAI-compatible key. Context window, max output and upstream list price per 1M tokens; open Details on any row for the full breakdown. For safety research and guardrail testing only.

upstream list price · USD per 1M tokens

Model catalogue: price per 1M tokens, context window and maximum output.
Model	Provider	Context	Max out	Price in / out
Claude Fable 5.1	Anthropic	1M	128K	$10.00in $50.00out
Anthropic's top tier above Opus, for the hardest reasoning and longest-horizon agentic work. ReasoningToolsVisionFiles1M context Input $10.00 Output $50.00 Cache read $0.25 Context 1,000,000 Max output 128,000 Released 2026-09-01 Cache write bills at 5× the input rate — caching is the cheapest way to run this tier.
Claude Opus 5.5	Anthropic	1M	128K	$4.00in $20.00out
Frontier-level performance for long-running agentic coding and knowledge work, at roughly half the price of Fable 5.1. ReasoningToolsVisionFiles1M context Input $4.00 Output $20.00 Cache read $0.20 Context 1,000,000 Max output 128,000 Released 2026-09-22 Cut from $5.00/$25.00 at release. Cache read is 5% of input, cache write 5× input.
Claude Sonnet 5.5	Anthropic	1M	128K	$2.00in $10.00out
Fast Claude model for everyday coding, agents and knowledge work — the flagship all-rounder of the family. ReasoningToolsVisionFiles1M context Input $2.00 Output $10.00 Cache read $0.10 Context 1,000,000 Max output 128,000 Released 2026-09-28 Full 1M window at half the Sonnet-4.6 price. The default RawLine pick for most work.
Claude Haiku 5.5	Anthropic	1M	128K	$0.10in $0.50out
Fast Claude model for responsive assistance, classification and lightweight agents — a full 1M window at the lowest rate here, tied with GPT-6 Luna. ReasoningToolsVisionFiles1M contextFast Input $0.10 Output $0.50 Cache read $0.01 Context 1,000,000 Max output 128,000 Released 2026-10-07 Rates shown cover requests up to 100K input tokens; longer ones fall into a higher tier — read it from `GET /v1/models`.
GPT-6 Astra	OpenAI	1.05M	128K	$10.00in $50.00out
OpenAI's most capable model for complex reasoning, coding, computer use, research and document creation, with reasoning effort up to max. ReasoningToolsVisionFiles1M contextComputer use Input $10.00 Output $50.00 Cache read $1.00 Context 1,050,000 Max output 128,000 Released 2026-09-04 Rates cover input up to 272K tokens; past that the long-context tier roughly doubles input and output.
GPT-6.1 Sol	OpenAI	1.05M	128K	$2.00in $10.00out
Near-Astra performance for complex coding, computer use and professional work, at a fifth of the price. The newest GPT generation. ReasoningToolsVisionFiles1M contextLatest Input $2.00 Output $10.00 Cache read $0.10 Context 1,050,000 Max output 128,000 Released 2026-09-29 Same 272K long-context tier split as Astra — the long tier roughly doubles both rates.
GPT-6 Luna	OpenAI	1.05M	128K	$0.10in $0.50out
Budget tier of the GPT-6 family — same context window and capabilities as Astra at a hundredth of the price. ReasoningToolsVisionFiles1M contextCheapest Input $0.10 Output $0.50 Cache read $0.01 Context 1,050,000 Max output 128,000 Released 2026-09-22 Halves the GPT-5.6 Luna rate. Best value for high-volume agent loops.
Gemini 3.1 Pro Preview	Google	1M	64K	$2.00in $12.00out
Reasoning-first Gemini preview for agentic coding and complex problem solving, with native audio and video input. ReasoningToolsVisionFilesAudioVideo1M contextPreview Input $2.00 Output $12.00 Cache read $0.20 Context 1,048,576 Max output 65,536 Released 2026-02-19 Rates cover prompts up to 200K tokens; past that Google bills $4.00/$18.00. Preview models can be retired without notice.
Gemini 3.8 Flash	Google	1M	64K	$0.75in $3.75out
Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents and enterprise workflows. ReasoningToolsVisionFilesAudioVideo1M contextFast Input $0.75 Output $3.75 Cache read $0.075 Context 1,048,576 Max output 65,536 Released 2026-09-02 Promo rate. $0.75/$3.75 holds through 31 Dec 2026, then reverts to $1.50/$7.50.
Grok 4.7	xAI	500K	500K	$2.00in $6.00out
xAI's frontier model for long-running agents, coding, knowledge work and visual projects. ReasoningToolsVisionFiles500K context Input $2.00 Output $6.00 Cache read $0.50 Context 500,000 Max output 500,000 Released 2026-09-21 Rates cover prompts up to 200K tokens; the long-context tier roughly doubles them. Max output equals the whole window.
Kimi K3	Moonshot	1M	1M	$3.00in $15.00out
Multimodal Kimi model with a 1M window and toggleable max-effort thinking for long-horizon agent work. ReasoningToolsVisionVideo1M contextOpen weights Input $3.00 Output $15.00 Cache read $0.30 Context 1,048,576 Max output 1,048,576 Released 2026-07-16 Open weights — self-host it when per-token cost matters more than the managed endpoint.
GLM-5.3	Z.ai	1M	128K	$1.40in $4.40out
Flagship GLM model for long-horizon coding, agents and complex project delivery, at the same rates as 5.2. ReasoningTools1M contextOpen weights Input $1.40 Output $4.40 Cache read $0.26 Context 1,000,000 Max output 131,072 Released 2026-08-14 Text only. For image input take GLM-5.3-Flash at $0.15/$0.50 — over nine times cheaper on input.
DeepSeek V4 Pro	DeepSeek	1M	384K	$0.66in $1.98out
Million-token context with thinking and non-thinking modes, switchable per request. ReasoningTools1M contextOpen weightsCheapest Input $0.66 Output $1.98 Cache read $0.022 Context 1,000,000 Max output 393,216 Released 2026-08-12 Off-peak rate. Peak hours (01:00–04:00 and 06:00–10:00 UTC) bill $1.32/$3.96 — double the rate above.
Qwen3.8 Max	Qwen / Alibaba	1M	128K	$2.00in $6.00out
2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding and long-horizon agentic workflows. ReasoningToolsVisionFilesVideo1M context Input $2.00 Output $6.00 Cache read $0.25 Context 1,000,000 Max output 131,072 Released 2026-08-03 The 3.8 generation dropped the old Max/Plus/Turbo naming. Qwen3.8-Flash runs this work at $0.15/$0.47.
Muse Spark 1.3	Meta	1M	128K	$1.25in $4.25out
Multimodal reasoning model from Meta for long-running agentic, multi-agent and coding workflows — the widest input range here, audio included. ReasoningToolsVisionFilesAudio1M context Input $1.25 Output $4.25 Cache read $0.15 Context 1,048,576 Max output 131,072 Released 2026-09-02 Meta publishes no first-party API — these are relay rates for the hosted weights.
Step-5 Preview	StepFun	1M	64K	$0.96in $2.74out
StepFun's flagship base model for coding and professional knowledge work, with native text, image and video input. ReasoningToolsVisionVideo1M contextPreview Input $0.96 Output $2.74 Cache read $0.048 Context 1,000,000 Max output 65,536 Released 2026-09-16 Preview model — expected to leave preview as the Step-5 family settles.

Nothing matches. Try another query, or pull the live list from GET /v1/models.

Prices are upstream list prices as of 11 Oct 2026, per 1M tokens in USD — what the labs charge. Several models also carry a long-context or off-peak tier the table doesn't show, so read the live per-token numbers from the API. Unpriced models answer 402 price_not_set — never served free.

Picking one

Hardest problems

Claude Opus 5.5 for reasoning-heavy work, GPT-6 Astra for the newest OpenAI generation, Claude Fable 5.1 above both.

Daily driver

Claude Sonnet 5.5 for a flagship all-rounder, GPT-6.1 Sol for coding and computer use at a fifth of Astra.

Volume work

GPT-6 Luna and Claude Haiku 5.5 — both $0.10/$0.50 with a full 1M window. DeepSeek V4 Pro runs cheaper off-peak.

Long context

Every model here holds 500K–1M tokens. Kimi K3 can also output a full 1M.

Self-host instead

GLM-5.3, DeepSeek V4 Pro and Kimi K3 ship open weights — run them yourself when per-token cost outweighs the managed endpoint.

Beyond text

Gemini 3.1 Pro and Muse Spark 1.3 take audio; Gemini and Step-5 take video. Text-only: GLM-5.3 and DeepSeek V4 Pro.

Details list and prices

This page is a static snapshot. The authoritative list — context length, max output, supported parameters and per-token pricing in USD — comes from the API:

curl https://app.rawline.dev/v1/models \
  -H "Authorization: Bearer rlk_YOUR_KEY"

A model whose providers all lack credentials is omitted from the response. See the API reference for the full field list and error codes.

Get an API key