NEWMEN
NEWMEN · TWO ENV VARS · $5 FREE

Cut your AI bill.

Atlas

Same quality, or it’s free.

Point your coding agent at Newmen and keep the models you use. You pay less than going direct — at least 5% off from your first call, climbing as Atlas ramps. Don’t like an answer? Thumbs-down it, full refund. You see the model that served every call and exactly what you saved.

Atlas speaks both OpenAI Chat Completions and Anthropic Messages on the wire. Codex CLI, Claude Code, Cursor, OpenAI / Anthropic SDKs, langchain, llamaindex — everything you already use just works.

export ANTHROPIC_BASE_URL="https://api.newmen.ai"
export ANTHROPIC_API_KEY="nm_live_..."

Claude Code calls https://api.newmen.ai/v1/messages

Sign in to insert your real key automatically.
Claim $5 in credits →How to migrate
  • Never more than going direct
  • Thumbs-down → full refund
  • No card to start · no subscription

Start in strict mode — same models, nothing rerouted, just cheaper — and let Atlas ramp when you’re ready. We train our optimization engine, never a model on your data: your prompts and code are never absorbed, sold, or shared.

Worked example · 30M tokens/mo · openai/gpt-5.5 pinned

$210/mo direct → $84/mo on Newmen

Your price vs going direct — verified per call. At least 5% off from call one, climbing as Atlas ramps. Don’t like a call? Thumbs-down it → full refund. See the full math →

Not theory — we ran a real coding agent through Newmen and built the same app twice. Read the proof: same apps, half the bill →

Saved this weeklive
Platform
two env varsMigration
5% from call oneSavings
thumbs-down refundQuality
served model per callTransparency
no card requiredFree models
engine, not a modelData
See all models →
01 — Every model, one key

Top by intelligence.

Pin any provider id you already use — openai/gpt-5.5, anthropic/claude-opus-4.7, meta-llama/llama-4-maverick — at or below the provider-direct rate. Atlas serves each call the cheapest way that holds quality, and you always see which model answered.

Top by intelligenceSee all 311+ models →
ProprietaryOpen weightsComposite of MMLU-Pro · HumanEval · GPQA · math

Everyone else says “trust us, the cheaper one is fine.” We prove it on your own traffic — you see the model that served each call, never pay more than direct, and get your money back if you don’t like it. Two env vars to switch.

02 — How Atlas works

Cheaper without quality loss.

Atlas serves each call the cheapest way that holds quality — and never charges you more than the provider would direct. You stay in control: pin any model in strict mode, or let Atlas optimize and ramp as you build confidence.

  • Atlas mode (default) — auto-optimize each call for the cheapest path that holds quality. At least 5% off from call one, climbing as it ramps on your traffic.
  • Strict mode — pin an exact model, sourced cheaper. The same model, pure pass-through, just less.
  • Verified — you see the served model and your price vs direct on every call, with a signed receipt you can reconcile against your provider bill.

Don’t like a call? Thumbs-down it → full refund, no questions. We only make money when you save money.

Qualityeval-gated accuracyCostadapters get cheaperSpeedoptimal routing latencyAtlasrouting

03 — Switch in seconds

Two env vars. Any tool.

Codex CLI, Claude Code, Cursor, the OpenAI / Anthropic SDKs in any language, langchain, llamaindex, OpenRouter clients — they all read the same env vars. Set them and ship.

export ANTHROPIC_BASE_URL="https://api.newmen.ai"
export ANTHROPIC_API_KEY="nm_live_..."

Claude Code calls https://api.newmen.ai/v1/messages

Sign in to insert your real key automatically.
Full migration guideBrowse all modelsAPI reference

Why Newmen

Never pay more than direct.

One key. Any model. We charge you the cheapest path that holds quality, never more than the original provider would direct, and credit you back automatically when a call disappoints.

At least 5% off, from call one

You never pay more than going direct, and you start saving on the very first request — climbing as Atlas ramps on your traffic. Every dollar saved shows up line-by-line in your console.

Thumbs-down → full refund

Don't like a call? Hit thumbs-down and we refund it in full, no questions asked. Your $5 of signup credits are real money on a real card — and free models cost nothing at all.

Two env vars to switch

Set OPENAI_BASE_URL or ANTHROPIC_BASE_URL plus a Newmen key. Codex CLI, Claude Code, Cursor, OpenAI / Anthropic SDKs, langchain, llamaindex — everything just works. No SDK rewrite. Existing model ids keep working.

The guarantee

If quality drops, the call isn’t billed.

Every call is covered: thumbs-down any response and it’s refunded in full, no questions asked. Want it automatic? Tag calls with operation_id, bind evaluators with a minimum score, and Atlas verifies each tagged operation against your live eval history — calls below the bar refund synchronously, before anyone reviews them. Calls that don’t pass never show up on your invoice.

  • Never more than going direct
  • Thumbs-down → full refund
  • No card to start · no subscription

Thumbs-down refunds are on every plan. Automatic eval-gated refunds are part of the Pro Reliability Loop, from $1,500 / month.

See how it works →

See it on your own prompts.

Paste 20–100 representative prompts. We’ll run them against your current model and against Atlas in parallel and show you the cost, latency, and (if you bind evaluators) eval-score deltas. Shareable result page you can forward to your team.

Try on your prompts →Browse all modelsSee pricing
  • Never more than going direct
  • Thumbs-down → full refund
  • No card to start · no subscription

Talk to a solutions engineer

Atlas is sold to teams who commit to meaningful production volume. That commitment unlocks the reliability loop.