Summarize with AI
Inception has released Mercury 3, a diffusion-based language model that runs at 1,068 tokens per second and has a 1M-token context window. It is available on the Inception API Platform and through OpenRouter, Baseten and Vercel.
Inception announced Mercury 3 on October 19, 2026, and calls it "the most capable diffusion-based LLM released to date." The company says Mercury 3 has more than 40x the total parameters of Mercury 1. It presents the release as evidence that the scaling laws behind autoregressive models also hold for text diffusion. According to Inception, Mercury 3 matches or beats GPT-6 Luna and Claude Haiku 5.5 across a range of benchmarks while running 4 to 8 times faster than them.
What it is
- Model type. A diffusion LLM. Inception calls it the flagship of its product line, alongside Mercury Voice and Mercury Decide.
- Speed. 1,068 tokens per second on NVIDIA GPUs.
- Context. 1M tokens, up from 260K for Mercury 2.5.
- Scale. More than 40x the total parameters of Mercury 1. Inception says scores on standard benchmarks rose 3x from Mercury 1 (mini) to Mercury 3.
- Use cases. Inception says early-access partners use it for browser-use agents, interactive education, real-time search agents and customer support.
- Compatibility. The API is OpenAI-compatible. Enterprise deployments offer dedicated capacity, autoscaling, compliance controls and configurable data retention.
- Availability. Inception API Platform, OpenRouter, Baseten and Vercel. New API users get 100 million free tokens.
Price
| Tokens | List price (per 1M) | Launch price, 50% off (per 1M) |
|---|---|---|
| Input | $0.20 | $0.10 |
| Output | $0.70 | $0.35 |
Cached input tokens cost 90% less than the input price.
Benchmarks
Inception published these scores. All three models were run at medium settings.
| Benchmark | Mercury 3 | GPT-6 Luna | Claude Haiku 5.5 |
|---|---|---|---|
| Output speed (tokens/s) | 1,068 | ~130 | ~240 |
| GDPval (Elo) | 1,237.2 | 1,204.0 | 1,225.0 |
| τ³-bench Banking | 30.24 | 23.00 | 13.06 |
| MultiChallenge | 60.35 | 51.49 | 52.48 |
| AA-Omniscience-Public (Accuracy) | 46.44 | 46.94 | 37.22 |
| IFBench | 76.84 | 61.82 | 54.65 |
| GPQA Diamond | 89.23 | 86.36 | 84.51 |
| AA-LCR v1.1 | 77.67 | 78.30 | 74.70 |
| MRCR v2 | 75.50 | 62.72 | 78.40 |
| SWE-Bench Atlas | 51.08 | 33.33 | 32.26 |
| DeepSWE v1 | 46.31 | 37.76 | 51.62 |
| BrowseComp+ | 55.18 | 50.50 | 45.78 |
| DeepSearchQA | 71.33 | 60.00 | 65.22 |
Mercury 3 leads on 8 of the 12 quality benchmarks. GPT-6 Luna leads on AA-Omniscience-Public and AA-LCR v1.1. Claude Haiku 5.5 leads on MRCR v2 and DeepSWE v1. These are the company's own numbers and have not been independently verified.
Sources
Did this land? Leave a reaction.
Discussion
0 commentsMembers can join the discussion
Become a member to comment and reply to other members.
Become a MemberNo comments yet
Be the first to start the discussion.