Quesma Blog
Field notes on coding agents, token economics, benchmarks and AI evaluation from the Quesma team.
- Posts
- 66 from 31 Oct 2023 to 1 Oct 2026
- Posting
- weekly (33 in the last year)
- Kind
- company blog
- Feed
- https://quesma.com/rss.xml
2026 21
- Your agent session transcripts are precious, keep them
- Token economics in Amsterdam: Inside the first Tokenomicon
- Hackathons in the vibe coding era: our token economics experiment
- GPT-6 Astra solves puzzles
- RTK reports huge token savings, but our cost benchmarks disagree
- OpenAI Codex pricing: the $270 PR a $200/month sub covers daily
- Gemini 3.7 Flash, Grok 4.6, GLM-5.3 and DeepSeek V4 Pro joined the frontier
- Claude Code pricing: same tokens, same model, up to 40x the price
- I wired 4 models together in Claude Code. One refused, and it backfired 4 ways on Terminal-Bench
- Quantization hurts knowledge nonlinearly - Qwen3.6 27B case study
- Kimi K3 is Open, Opus 5 is Good, DeepSeek V4 Flash is Cheap: LLMs on Baba Is You
- A lesson about retries, hidden in the DeepSeek-V4 paper
- Do Qwen3.6 27B quantizations break the pelican?
- I burned all my tokens researching how to save tokens
- Baba Is Solved by Fable 5 and GPT-5.6 Sol, but at what cost?
- Tokenflation: When “Hi” triggers 33 tool calls
- Compare harnesses, not models: Blitzy vs. GPT-5.4 on SWE-Bench Pro
- Reverse engineering River Raid with Claude, Ghidra, and MCP
- Benchmarking OpenTelemetry: can AI trace your failed login?
- Vibe coding needs git blame
- How 2025 took AI from party tricks to production tools
2025 19
- Migrating CompileBench to Harbor: standardizing AI agent evals
- Antigravity feels heavy and Claude Skills are light
- A postmortem on our $2.5 million database gateway: lessons from pilot purgatory
- The security paradox of local LLMs
- CompileBench: can AI compile 22-year-old code?
- Tau² benchmark: how a prompt rewrite boosted GPT-5-mini by 22%
- 60-second Linux analysis - with Nix and LLMs
- Tau²: from LLM benchmark to blueprint for testing AI agents
- Observability in Go: what real engineers are saying in 2025
- Claude Code + OpenTelemetry + Grafana: a guide to tracking usage and limits
- 5 Grafana in Docker examples to get started with metrics, logs, and traces
- How does the Data Council conference keep finding future billion-dollar companies?
- Don’t let Apache Iceberg sink your analytics: practical limitations in 2025
- Highlights from the Iceberg Summit 2025
- February 2025 newsletter
- Lessons from the pre-LLM AI in observability: anomaly detection and AIOps vs P99
- ClickHouse Cloud pricing change in January 2025: a price hike with many tweaks
- January 2025 newsletter
- Team update
2024 24
- December 2024 newsletter
- Quesma database gateway
- November 2024 newsletter
- We are announcing our first public release at KubeCon!
- The most successful open-source fork, worth $6 billion
- October 2024 newsletter
- Hydrolix and Quesma technical partnership announcement
- SQL from a programming language perspective — Part II
- SQL from a programming language perspective — Part I
- Understanding Elasticsearch pricing
- Getting started with Quesma in 3 simple steps
- The pancake SQL pattern: combine your SQLs into one for 50x better performance
- Schema migrations: pitfalls and risks
- The importance of design partners for product development
- Maximizing Elasticsearch’s performance and cost efficiency
- What SQL could learn from Elasticsearch Query DSL
- How Quesma can help Kibana users reduce Elasticsearch costs
- Elasticsearch and OpenSearch schema rough edges
- OpenSearch vs. Elasticsearch: a fork in the road
- Kibana vs. Grafana: choosing the right data visualization tool
- Kibana on ClickHouse data with Quesma - dashboarding
- Kibana on ClickHouse data with Quesma - first working prototype
- Full-text search X times faster: inverted index vs. SQL OLAP
- Best tool for the job