CL-SN55 941 —
GPT-61S 885 —
GPT-56T 861 —
GPT-6S 825 —
CL-OP55X 818 -0.1%
MUSE-SPK 793 —
QWEN-38X 779 —
GPT-6A 778 —
CL-OP5H 711 -0.1%
CL-OP5X 709 -0.1%
KIMI-K3X 699 -0.1%
CL-OP46H 694 -0.1%
CL-FAB5H 692 -0.1%
CL-OP55H 692 -0.1%
CL-OP47H 683 -0.1%
GEM-37FH 648 -0.2%
GEM-38FH 633 -0.2%
INKL 531 —
CL-OP47 529 -0.2%
GPT-56S 529 -5.9%
GEM-3P 493 —
CL-OP46 488 —
CL-OP48 481 —
GEM-31P 453 -0.2%
GPT-6L 436 —
CL-SN55 941 —
GPT-61S 885 —
GPT-56T 861 —
GPT-6S 825 —
CL-OP55X 818 -0.1%
MUSE-SPK 793 —
QWEN-38X 779 —
GPT-6A 778 —
CL-OP5H 711 -0.1%
CL-OP5X 709 -0.1%
KIMI-K3X 699 -0.1%
CL-OP46H 694 -0.1%
CL-FAB5H 692 -0.1%
CL-OP55H 692 -0.1%
CL-OP47H 683 -0.1%
GEM-37FH 648 -0.2%
GEM-38FH 633 -0.2%
INKL 531 —
CL-OP47 529 -0.2%
GPT-56S 529 -5.9%
GEM-3P 493 —
CL-OP46 488 —
CL-OP48 481 —
GEM-31P 453 -0.2%
GPT-6L 436 —
← Back to feed

Pareto: Unbiased Ships a Blended AI Model That Runs Every Request Through Multiple Frontier Models at $2.50 per Million Tokens

Unbiased, a startup focused on AI API infrastructure, launched Pareto 26.9 on September 17. Pareto is not a single model: it runs every request against a blend of frontier and open-source models simultaneously, synthesizes one answer, and returns it through a single OpenAI-compatible API call. Pricing is $2.50 per million input tokens and $7.50 per million output tokens on OpenRouter, with a 262,144-token context window.

Architecture: ensemble, not router

Unbiased draws a sharp distinction between Pareto and model routers. A router inspects each prompt, estimates complexity, and dispatches to one model. Pareto runs the same blend on every request, hard or easy. The company argues this matters at scale for a specific reason: routers break prompt caching when they switch models between turns, erasing the cost savings that make long conversations economical. Pareto never switches models mid-conversation, so prompt cache hits remain intact across extended trajectories.

The model composition is not static. Unbiased publishes release notes when the blend changes, and the model card is versioned by release date. The September 2026 release (Pareto 26.9) benchmarks against Fable 5.1, GPT-6 Astra, and DeepSeek V4.1 Flash.

Capabilities

Pareto accepts text and image inputs and returns text. It supports function calling via the tools parameter but does not enforce structured JSON output through response_format. Completion tokens are capped at 131,072. The model is available on OpenRouter and is also integrated into Cloudflare AI Workers.

Signup on the Unbiased platform is gated: every account is reviewed manually before a key is issued. Volume users can request an onboarding call.

Evaluation approach

Unbiased publishes an open evaluation harness called pareto-evals that runs head-to-head comparisons against any OpenAI-compatible endpoint. Accuracy and cost per task are measured simultaneously, with task costs read from the API response itself rather than estimated. No composite score from an independent evaluator has been published for the 26.9 release.

Pricing context

At $2.50/$7.50, Pareto undercuts most frontier single models while claiming comparable quality. Claude Fable 5.1 at Anthropic’s list rates runs considerably more expensive per token. The trade-off is the function calling constraint and the absence of response_format support, which rules Pareto out for strict JSON output use cases without wrapper logic.