MModelSite·2d agoCall DeepSeek, Qwen and GLM with one OpenAI-compatible API keySame OpenAI SDK, one endpoint: change the model name and keep your code.2 min read·modelsite.hashnode.dev
KKris Holy·2d agoAI Inference Platforms Benchmark 2026: Fireworks, Together, Baseten, Modal, DeepInfra, Z.ai and KelnWe replayed real agent sessions on DeepSeek V4.1 Flash and GLM-5.3 Flash and compared speed, reliability, prompt caching and cost.10 min read·keln.hashnode.dev
Xxiun xiao·2d agoDeepSeek V4.1 Flash vs Claude Haiku 5.5: a 17-task benchmark of the two cheapest tiersThe two cheapest tiers on the same 17 coding, extraction and reasoning tasks, three runs each: pass rate, latency and cost per 1,000 tasks.6 min read·xiunxiao.hashnode.dev
KKarmaBox·2d agoA 284B Model on a 64GB PC Has to Beat $0.00086 a Task. Here's the Mathllama-modchip fits DeepSeek-V4-Flash into 64GB of RAM with no GPU. We priced that box against per-token billing, with every assumption shown.20 min read·karmabox.hashnode.dev
李李弘基·5d agoDeepSeek Superagent: Don't Chat in One Window. Hand Work to a Repo Worker.Check the environment, send one small GitHub task, and review the result before expanding its responsibilities.13 min read·kd-agentic.hashnode.dev
JJakson Tate·5d agoDeepSeek DeepInfer: Production Inference Architecture, DSpark Optimization & Bare-Metal GPU DeploymentAn SRE guide to DeepSeek-V4.1-Flash inference, SWA bounded replay, speculative decoding, and production serving requirements.6 min read·servermo.hashnode.dev
Mming·Nov 3Does It Save Tokens? TokenEconomizer on DeepSeek HarnessI ported TokenEconomizer to DeepSeek Harness and ran 12 A/B sessions. Total tokens: 209,263 with the plugin, 208,239 without, a 0.5% gap.8 min read·bigai.hashnode.dev
999 sunny·Nov 1Native multimodal still allows a separate encoderNative multimodal does not mean dropping separate vision encoders; DeepSeek-V4.1-Flash still uses ViT + projector, and the real goal is an updatable world state.7 min read·sunny99.hashnode.dev
MMagnus·Nov 1An open-source desktop agent that follows DeepSeek's super-agent design (and what we left out)Disclosure: I maintain AnyModelBot. This is an unofficial project, not affiliated with DeepSeek. Most chatbots wait for you to speak and stop when they answer. Plenty of real work is not a question an6 min read·modelflare.hashnode.dev
CCloudGPU·Oct 29Roo Code configuration profiles for DeepSeek, Kimi and MiniMax on one OpenAI-compatible endpointOne base URL and one key, a profile per model ID, a profile per mode, and a tool-call check for each6 min read·cloudgpu.hashnode.dev
