model roundup

DeepSeek 4

8 items · started 2026-06-22 · ongoing (last activity 2026-06-25)

  1. There is an adversarial relationship between developers and big model labs. Model labs charged developers higher API prices to subsidize their own agent harness offerings.

  2. Inferize is building highly optimized, elastic inference for AI workloads. Ridiculously fast, efficient LLM serving that scales with demand.

  3. LLMs seem to love certain languages (Python, Bash, etc.), but they all seem to struggle with Lisp (e.g. Racket or Emacs Lisp).

  4. ds4 - Mixed NVFP4 serving of DeepSeek V4 Flash on the NVIDIA Spark family (GB10) ⚠️ This GitHub repository is for archival / mirror purposes only. Active development happens at git.kokoham.com/sleepy/ds4-nvfp4-spark.

  5. Microsoft is reportedly considering introducing a fine-tuned version of the Chinese open-source model DeepSeek V4 into its enterprise artificial intelligence (AI) tool Copilot Cowork, as a lower-cost alternative to models from OpenAI and A…

  6. Which option gives the most actual Opus 4.8 usage volume: Kiro Pro, Claude Pro or something else? My monthly budget is $30.

  7. We present a preview version of DeepSeek-V4 series, including two strong Mixture-of-Experts (MoE) language models -- DeepSeek-V4-Pro with 1.6T parameters (49B activated) and DeepSeek-V4-Flash with 284B parameters (13B activated) -- both su…

  8. I’ve been using GLM 5.2 with Claude Code through its Anthropic-compatible API endpoint. I’ve tested it on various projects, including but not limited to database development, backend payment API work, backend and frontend debugging, Larave…

← all threads