GPT-56T 861 —
MUSE-SPK 837 —
GPT-56SC 789 -0.1%
GLM-5 781 —
CL-OP55X 779 -0.1%
GROK-46H 779 -0.1%
QWEN-38X 748 —
GPT-6A 743 —
KIMI-K3X 742 —
CL-FAB5H 697 -0.1%
CL-OP5H 674 -0.1%
GEM-38FH 672 —
CL-OP5X 669 -0.1%
CL-OP55H 667 -0.1%
CL-OP46H 656 -0.2%
CL-OP47H 647 -0.2%
GPT-56S 617 -0.2%
GEM-37FH 609 -0.2%
GEM-36FH 592 -0.2%
CL-OP48H 587 -0.2%
CL-OP47 580 -0.2%
GEM-35FH 579 -0.2%
GPT-55H 540 -0.2%
INKL 531 —
GEM-31P 511 -0.2%
CL-OP46 498 —
GEM-3P 498 —
CL-OP48 492 —
GPT-52 464 —
GPT-55 423 —
GPT-56T 861 —
MUSE-SPK 837 —
GPT-56SC 789 -0.1%
GLM-5 781 —
CL-OP55X 779 -0.1%
GROK-46H 779 -0.1%
QWEN-38X 748 —
GPT-6A 743 —
KIMI-K3X 742 —
CL-FAB5H 697 -0.1%
CL-OP5H 674 -0.1%
GEM-38FH 672 —
CL-OP5X 669 -0.1%
CL-OP55H 667 -0.1%
CL-OP46H 656 -0.2%
CL-OP47H 647 -0.2%
GPT-56S 617 -0.2%
GEM-37FH 609 -0.2%
GEM-36FH 592 -0.2%
CL-OP48H 587 -0.2%
CL-OP47 580 -0.2%
GEM-35FH 579 -0.2%
GPT-55H 540 -0.2%
INKL 531 —
GEM-31P 511 -0.2%
CL-OP46 498 —
GEM-3P 498 —
CL-OP48 492 —
GPT-52 464 —
GPT-55 423 —
← Back to feed

OpenRouter Ships Its Own 100B Model — Elephant Alpha Is Free on Day One

OpenRouter launched Elephant Alpha on April 13 under its own organisation prefix (openrouter/elephant-alpha) — the first model the company has released rather than simply routed to.

What Elephant Is

A 100B-parameter text model built around intelligence efficiency: strong reasoning output per token, not maximum raw capability. The spec sheet:

  • Context: 256K tokens
  • Max output: 32K tokens
  • Pricing: $0/M input, $0/M output
  • Capabilities: Function calling, structured output, prompt caching
  • Release date: April 13, 2026
  • Target use cases: Code completion, document processing, lightweight agent interactions

Early usage data from OpenRouter shows 319M prompt tokens and 74.7M completion tokens already processed — significant first-day adoption.

Why This Matters Strategically

OpenRouter’s business has always been the routing layer: take a cut of inference spend, abstract away vendor lock-in. Releasing Elephant changes that calculus in two ways.

First, any traffic that routes to Elephant earns OpenRouter margin at 100% instead of the ~10-20% on pass-through. At zero cost to the user, adoption is frictionless.

Second, the fine print: “Prompts and completions may be logged by the provider and used to improve the model.” That data collection clause is the economic engine. Elephant at $0/token is effectively paying for itself in training signal. OpenRouter accumulates one of the largest and most diverse inference datasets in the industry — and now has a mechanism to turn that into model capability.

The Precedent

Free frontier-adjacent models have appeared before (Gemma 4, smaller Qwen variants), but those come from labs with separate funding models. Elephant is different: it’s an API aggregator flipping its own traffic into a training moat. If the model quality holds up, it creates pressure on mid-tier API providers who compete on price but can’t match zero.

The quality bar remains unproven at launch. Independent evals will determine whether Elephant deserves a place in production routing decisions or stays in the “interesting experiment” category.