ModelsQwen3.7 Max Guide: The Short-Lived 3.7 Flagship
Qwen3.7 Max led Alibaba's line-up from May 2026 until Qwen3.8 Max arrived in August. What it does, which snapshot you are calling, and whether it still earns its price.
ReadNotes on models, agents, and what inference costs
Practical writing for developers building with large language models — how they work, how to pick one, and how to keep the bill predictable.
Models
ModelsQwen3.7 Plus: The Mid-Tier Qwen Built for Screen-Reading Agents
Qwen3.7 Plus pairs a 1M window with vision aimed at GUI agents, at a fraction of Max pricing. What it claims, what it costs, and when an open 27B is the better call.
Read
ModelsQwen3.8 2.4T A95B: The Open Weights Behind Qwen3.8 Max
Qwen published 2.4T-parameter MoE weights with 95B active, text-only and always thinking. What is in them, what the licence allows, and why the card scores are not theirs.
ReadModelsQwen3.8 27B: A Dense Apache-Licensed Model With Agent Ambitions
Qwen3.8 27B is a dense vision-language model under Apache 2.0 with vendor scores close to far larger models. What it claims, how providers serve it, and how to run it.
Read
ModelsQwen3.8 Flash Guide: The Cheap Default in the Qwen3.8 Line
Qwen3.8 Flash pairs a 1M window and a thinking toggle with a low per-token price. What is documented, what the cache pricing does, and when to step up to Max.
ReadModelsQwen3.8 Max (0902): The Snapshot, the Model and the Price
Qwen3.8 Max 0902 is a dated snapshot of Alibaba's 2.4T flagship. What changed, what the open-weight sibling reveals about it, and how to pin and test it.
Read
ModelsQwen3.8 Max Prime: When Paying Double for Speed Makes Sense
Qwen3.8 Max Prime is a higher-throughput tier of Qwen3.8 Max at twice the price. What is known, what speed buys, and how to measure it before you pay for it.
ReadModelsQwen3.8 Omni Flash: Audio and Video In, Text Out
Qwen3.8 Omni Flash takes audio, video, images and text at the same price as Flash. What native audio understanding replaces, and how to test it on real recordings.
ReadModelsReka Edge Guide: A 7B Vision Model Built for Devices
Reka Edge is a 7B vision-language model aimed at Jetson boards, Macs and phones. Its token efficiency, its benchmark table, and the licence clause to read first.
Read
ModelsReka Flash 3 Guide: A 21B Apache Reasoning Model, Revisited
Reka Flash 3 is a 21B reasoning model from March 2025 under Apache 2.0. What it does well, how budget forcing works, and whether it still earns a place in 2026.
ReadModelsRelace Apply 3: The Small Model That Merges Your Agent Edits
Relace Apply 3 is not a chat model. It takes a lazy edit snippet and an original file and returns the merged file at high speed. What it does and when it pays off.
ReadModelsRelace Search: An RL-Trained Subagent for Finding the Right Files
Relace Search runs parallel grep and file-view calls to hand your main agent a short list of relevant files. How it was trained, what it saves, and what it costs.
ReadShowing 289–300 of 731 articles