Fireworks · Blogs

Blogs listings from Fireworks on TechiSeek. 148 public listings.

• Post-training Kimi K3 with Harvey for long-horizon legal work (Fireworks)
  Harvey recently announced its first model, Tenet, post-trained in collaboration with Fireworks for long-horizon legal work.
• FireAttention V4: Industry-Leading Latency and Cost Efficiency with FP4 (Fireworks)
  Today, we’re announcing we've achieved industry-leading speeds of >250 tokens/second on NVIDIA B200 GPUs using our latest FireAttention V4…
• Partnering with Meta: Bringing Llama 3.2 to Fireworks for Fine-Tuning and Inference (Fireworks)
  Access the latest Llama 3.2 models from Meta on Fireworks today.
• Fireworks.ai: Fast, Affordable, Customizable Gen AI Platform (Fireworks)
  Fireworks.ai releases the fast, affordable, and customizable Fireworks GenAI Platform.
• The frontier isn’t a model. It’s a router. (Fireworks)
  The best single model gets 74.1% at $6.52.
• Optimizing MiniMax M3 Sparse Attention on NVIDIA Blackwell (Fireworks)
  Fireworks built a KV-stationary sparse-attention kernel for MiniMax M3 on NVIDIA Blackwell (SM100), reaching ~980 TFLOP/s: 1.9–2.4× a query…
• How I shipped a month of engineering work in four days with GLM 5.2 Fast (Fireworks)
  A Fireworks engineer used GLM 5.2 Fast on Claude Code to ship a GPU-scheduler feature scoped at a full engineer-month.
• DeepSeek v3 and R1 Model Architecture: Why it's powerful and economical (Fireworks)
  This article introduces the architecture of DeepSeek v3 and DeepSeek R1, which shares exactly the same architecture.
• Every byte counts: ARCv3 and the case for cross-region RL (Fireworks)
  Fireworks’ ARCv3 delivers lossless BF16 weight updates nearly 50% smaller than ARCv2, making cross-region reinforcement learning more pract…
• Multi-Query Attention is All You Need (Fireworks)
  The newest Mistral AI MoE model, Mixtral 8x7B, is available on the Fireworks platform in both base and instruction-tuned variants.
• Partnering with Meta to bring Llama 3 to Firework’s inference and fine-tuning (Fireworks)
  We are pleased to announce the availability of the open-source Llama 3 8 B and 70 B models with 8 k context, served from our blazing fast i…
• Qwen 3 on Fireworks: Controllable Chain-of-Thought and Tool Calling at Frontier Scale (Fireworks)
  Until now, open-source LLMs forced a choice: show the chain of thought or call tools deterministically.
• LLM Eval Driven Development with Claude Code (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• Fine-Tuning DeepSeek v3 & R1 to optimize quality, latency, & cost (Fireworks)
  Customization of DeepSeek R1 & V3, through Quantization Aware Fine Tuning, is now available as part of Fireworks https://fireworks.ai/blog/…
• The Benchmark Gap: What It Takes to Ship Kimi K2.5 (Fireworks)
  Quality is the most important characteristic of a frontier model, and we at Fireworks are committed to getting this right.
• 3D FireOptimizer: Automating the Multi-Dimensional Tradeoffs in LLM Serving (Fireworks)
  Fireworks 3D FireOptimizer automatically searches through thousands of options to find configurations that achieve the optimal quality, thr…
• How Enterprises are using Multimodal Models in production with Fireworks (Fireworks)
  Major enterprises use Fireworks to process unstructured data in real-time, at higher accuracy, 100x lower cost and 1.5x faster speed than a…
• FireAttention V2: 12x faster to make Long Contexts practical for Online Inference (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• Build powerful agents on OSS models with Blazing Fast Inference on Fireworks (Fireworks)
  Fireworks delivers the industry's fastest, low-latency inference for top open-source models like Kimi K2.5, Deepseek V3.2, and GLM 4.7, as…
• Kimi K2.7 Code on Fireworks: Better Agents, Lower Cost per Task, Available Day-0 (Fireworks)
  Fireworks is launching Day-0 support for Kimi K2.7 Code, the latest model in Moonshot’s K2 coding series.
• The Fine-Tuning Bottleneck Isn't the Algorithm (Fireworks)
  The Fine-Tuning Bottleneck Isn't the Algorithm. Teams default to chasing the latest training algorithm, but the real bottlenecks are integr…
• Using Model-as-a-Judge for Reward in Reinforcement Finetuning (Fireworks)
  In domains that are inherently challenging to quantify, such as creative writing, we demonstrate that leveraging a superior large language…
• Introducing The Specialized Intelligence Index (Fireworks)
  Fireworks introduces the Specialized Intelligence Index, a one-stop destination for real-work benchmarks across industries.
• Self-Improving Agents, Powered by Your Evals (Fireworks)
  Eval Protocol is a unified eval interface that powers both prompt optimization and RL on the same evaluation function.
• Fireworks Now Supports NVIDIA NIM Deployments for Blazing AI Inference (Fireworks)
  Today, we’re pleased to announce that Fireworks supports NVIDIA NIM microservices, part of the NVIDIA AI Enterprise software platform, maki…
• Best LLMs for coding: 2026 roundup (Fireworks)
  Compare GPT-5.5, Claude Opus 4.7, DeepSeek V4-Pro, Kimi K2.6, and 7 more LLMs for coding. Artificial Analysis benchmarks, Fireworks pricing…
• Code Generation with Large Language Models - Fireworks Take (Fireworks)
  “Programming is the art of telling another human being what one wants the computer to do. We should continually strive to transform every a…
• Announcing Embeddings and Reranking On Fireworks (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• Multi-LoRA: Personalize AI at scale and deliver the best experience for each customer and use case (Fireworks)
  At Fireworks, we know how important it is for you to deliver the best product experiences to your users.
• Run bulk async workloads with Fireworks Batch API (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• Fireworks Raises the Quality Bar with Function Calling Model and API Release (Fireworks)
  Since rolling out the Fireworks platform earlier, we've been delighted to be part of the development journey of thousands of developers.
• Fireworks Secures $1.5 Billion in Series D Funding (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• Kimi K3 on Fireworks: Frontier Intelligence You Can Own (Fireworks)
  Kimi K3 on Fireworks brings frontier-level intelligence to an open-weight 2.8-trillion parameter model, delivering matching performance to…
• Best Open Source LLMs of May 2026: We Reviewed 7 Models (Fireworks)
  We tested 7 open source LLMs including DeepSeek v3.2, Kimi K2.5, and Qwen3 VL. Compare benchmarks, costs, and performance to pick the right…
• NVIDIA Nemotron 3 Ultra is live on Fireworks, day zero (Fireworks)
  Run NVIDIA Nemotron 3 Ultra on Fireworks. Optimize autonomous agents with 1M context and frontier reasoning.
• Three projects, one platform: A developer's winning streak with Fireworks (Fireworks)
  The project tackles a key issue in developer tools: ensuring reliable technical documentation.
• Doomed to Code: How we Teamed Up with Fireworks at MistralAI Hackathon to Conquer the Shores of Hell (Fireworks)
  A blog about how a bunch of hackers joined to make an LLM play DOOM, a PC game.
• Unlock Your Tools: Fireworks Adds OpenAI-Response API with MCP Support (Beta) (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• 20x faster Whisper than OpenAI - Fireworks audio transcribes 1 hour in 4 seconds (Fireworks)
  Fireworks now supports audio transcription, with a fast and comprehensive audio model. Transcribe 1 hour of audio in 4 seconds and use addi…
• Best Open Source LLMs in 2026: We Reviewed 7 Models (Fireworks)
  Compare Kimi K3 with GLM 5.2 and other leading LLMs using current benchmark results and deployment options, with weight and license status…
• Own Your AI: Fireworks Training Preview (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• DeepSeek V3.1 now on Fireworks! (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…
• Supervised Fine-Tuning (SFT) with LoRA on Fireworks: Tutorial (Fireworks)
  Supervised Fine-Tuning (SFT) is critical for adapting general-purpose Large Language Models (LLMs) to domain-specific tasks, significantly…
• Faster, more efficient DeepSeek on the Fireworks Developer Cloud (Fireworks)
  Discover how Fireworks Developer Cloud accelerates AI innovation with faster, optimized DeepSeek R1 deployments.
• Fireworks Platform Spring 2024 Updates (Fireworks)
  Fireworks launches faster models, higher rate limits, cheaper pricing, post-paid billing, and a new Business tier.
• FireOptimizer: Customizing latency and quality for your production inference workload (Fireworks)
  Introducing FireOptimizer, our adaptation engine designed to customize and enhance AI model performance for your unique use cases.
• Fireworks launches fine-tuning service - Rapidly iterate on quality and scale to production through (Fireworks)
  •Fine-tuning service launch Fireworks is offering an affordable LoRA fine-tuning service for 10 models, including Mixtral.
• with Stability’s API Powered by Fireworks (Fireworks)
  Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with…