Mercury 3
A diffusion-model-based generative AI model focused on scalable deployment. Through engineering optimizations such as few-step sampling, it reduces inference costs and achieves a commercially acceptable balance between throughput and latency.
Timeline (last 90 days)
Inception's Mercury 3 uses diffusion-based generation, with the vendor reporting 1,068 tokens per second and leading comparison models on 8 of 12 internal tests, but these are not independent benchmark conclusions.
Mercury 3 is not an end-user chat application but provides model capabilities in the form of an API, targeting development teams looking to replace their existing inference backend.
The Maker of Mercury 3 is Volodymyr Kuleshov, and the product is listed under three tags: API, Artificial Intelligence, and Developer Tools.
Mercury 3 is a diffusion language model launched on Product Hunt, which it officially calls the most capable diffusion LLM to date.
The Product Hunt page shows Mercury 3 received 94 votes and 2 comments, ranking 8th for the day.
It is officially claimed that Mercury 3 surpasses GPT-6 Luna and Claude Haiku 5.5 on multiple benchmarks.
Mercury was released on October 19, adopting a diffusion-style text generation approach, with context extended to 1 million tokens and standard input and output priced at $0.2 and $0.7 per million.
Mercury 3 is available only through Inception's Nvidia GPU API, with no open weights and no quantized versions.
Inception's Mercury 3 uses a diffusion architecture, filling multiple tokens at once, reaching 1,068 tokens per second on Nvidia GPUs, and has a 1 million token context window.
The AI News briefing said these three models "will not enter hyperscale use for the time being."
21 more timeline events
All Facts (20)
Inception's Mercury 3 uses diffusion-based generation, with the vendor reporting 1,068 tokens per second and leading comparison models on 8 of 12 internal tests, but these are not independent benchmark conclusions.
50%UnverifiedMercury 3 is not an end-user chat application but provides model capabilities in the form of an API, targeting development teams looking to replace their existing inference backend.
50%UnverifiedThe Maker of Mercury 3 is Volodymyr Kuleshov, and the product is listed under three tags: API, Artificial Intelligence, and Developer Tools.
50%UnverifiedThe Product Hunt page shows Mercury 3 received 94 votes and 2 comments, ranking 8th for the day.
50%UnverifiedIt is officially claimed that Mercury 3 surpasses GPT-6 Luna and Claude Haiku 5.5 on multiple benchmarks.
50%UnverifiedInception's Mercury 3 uses a diffusion architecture, filling multiple tokens at once, reaching 1,068 tokens per second on Nvidia GPUs, and has a 1 million token context window.
50%UnverifiedMercury 3 is available only through Inception's Nvidia GPU API, with no open weights and no quantized versions.
50%UnverifiedThe briefing did not disclose the parameter sizes, benchmark scores, or pricing strategies of these three models.
50%UnverifiedBefore the specific specifications of the three models are announced, it is advisable to wait and see, first confirming context length, toolchain compatibility, and pricing competitiveness.
50%UnverifiedThe specific background of Mercury 3 was not elaborated in the briefing.
50%UnverifiedThe AI News briefing said these three models "will not enter hyperscale use for the time being."
50%UnverifiedInception Labs released Mercury 3, a diffusion LLM, claiming it surpasses GPT-6 Luna and Claude Haiku 5.5 on benchmarks and runs 4 to 8 times faster; generation speed reaches 1,068 tokens/second on NVIDIA GPUs, and the context window expands from 260K to 1 million tokens.
50%UnverifiedOfficial data shows that Mercury 3 reaches an inference speed of 1,068 tokens per second.
50%UnverifiedAccording to Inception's own published comparison table, Mercury 3 leads GPT-6 Luna and Claude Haiku 5.5 on 8 of 12 benchmarks.
50%UnverifiedNew architectures typically face adaptation issues with toolchains, fine-tuning methods, and deployment frameworks, and information on whether Mercury 3 is compatible with mainstream inference frameworks and fine-tuning paradigms is currently limited.
50%UnverifiedWhen calling Mercury 3 on the AI Gateway, you need to set the environment variable AI_GATEWAY_API_KEY and specify the model name as inception/mercury-3.
50%UnverifiedInception claims that Mercury 3 achieved "strong results" in benchmarks, but the original text did not disclose the specific comparison models and score details.
50%UnverifiedTo use Mercury 3 in coding agents such as Claude Code, Codex, and Cursor, you need to install the latest version of the Vercel CLI and run npx vercel ai-gateway setup, then select inception/mercury-3 in the agent.
50%UnverifiedMercury 3 offers a context window of 1 million tokens.
50%UnverifiedAccording to figures published by Inception, Mercury 3 generates at over 1000 tokens per second.
50%