AI Weekly Malaysia

Anthropic introduced Claude Sonnet 5.5, the second model in the Claude 5.5 family, claiming 30%+ faster output and up to 30% lower cost per task than Sonnet 5 at unchanged list pricing of $2 per million input tokens, $10 per million output tokens, and $0.20 per million cache reads. It scores 70.6% on Terminal-Bench 4.0 versus Sonnet 5's 10.3%, comes within two points of Opus 5.5 on GDPval-AA, and is the first Sonnet model to ship with cyber safeguards and fallbacks; Haiku 5.5 is promised in the coming weeks. The Hacker News thread drew 390 points and 254 comments.

Why: If your coding agent or document pipeline defaults to Opus 5.5, this is a concrete reason to re-test model routing: Sonnet 5.5 claims 70.6% on Terminal-Bench 4.0 (the table lists Opus 5.5 at 66.4%, with a footnote) at $2/$10 per million tokens and 30%+ faster generation, so the cheaper model may now win on well-scoped bug fixes and slide/spreadsheet generation. Note these are Anthropic's own benchmark and cost figures — the 10.3% to 70.6% jump is large enough that you should run your own repo tasks through both before switching a default. Also flag the new cyber safeguards on a Sonnet-tier model: Anthropic says routine software development is unaffected, but anything security-adjacent you route through Sonnet may now hit fallbacks. For teams billing API usage in USD against MYR budgets, the token-efficiency claim (same per-token price, up to 30% fewer tokens per task) is the number to verify on your own workload.

OpenAI's DevDay 2026 recap lists 20+ announcements across ChatGPT, Codex, and its models, headlined by Dots (always-on agents on Pro and Business Premium in 'eligible markets', with Enterprise/Edu/Healthcare beta off by default), GPT-6.1 Sol (claimed near-Astra performance on agentic coding and computer use at one-fifth of Astra's standard input and output token prices), and an Ultrafast speed tier at 300 tokens/second (up to 8x faster in Codex, 6x in the API), with GPT-6 Astra Ultrafast available today in the API and on Pro 500/Enterprise plans. It also opens ChatGPT as a surface for plugins and native developer experiences, citing 1.2B weekly users. No independent benchmarks, latency numbers under load, or regional availability list are provided.

Why: Two concrete decisions hinge on this: if your token bill is currently the constraint on Astra-class agentic coding, GPT-6.1 Sol is pitched at 1/5 of Astra's standard input/output price, so it is worth benchmarking against your own evals before renewing spend. But only GPT-6 Astra Ultrafast (300 tok/s) is available today, and only via the API or Pro 500/Enterprise plans; GPT-6.1 Sol Ultrafast is 'coming soon', so don't commit a latency-sensitive product roadmap to it. Dots is off by default for Enterprise/Edu/Healthcare and restricted to unspecified 'eligible markets', so whether it is usable from Malaysia is not stated in this text and needs checking directly.

Google DeepMind introduced Gemini 4 Argon, claiming first place on 13 of 19 benchmarks against GPT-6 Astra and Claude Opus 5.5, with a 1M-token output limit via the new Long Decode Continuation API feature. Standard pricing is $4/$20 per 1M input/output tokens, with a 50% introductory discount to $2/$10 and 95% off cached input. Access is limited to government users and trusted cyber defenders in the Fairwind Program, with broader developer, enterprise, and consumer access promised later.

Why: The actionable details are gated: Argon is not generally available, and the 1M output is delivered via Long Decode Continuation, which pauses and resumes responses across calls, while Vals lists 262K max output. Don't re-architect around 1M single-call output yet; if you evaluate it later, compare the $4/$20 standard or $2/$10 intro pricing against your current model, and note cached input is 95% off.

A blog post on insufferable.dev argues the competitive dynamic between Western and Chinese AI labs has flipped: instead of Western labs accusing Chinese labs of distilling their models, Western labs are now quietly adopting Chinese inference optimizations. It cites DeepSeek's KV cache work — MLA at roughly 15x compression, then Compressed Sparse Attention and Heavily Compressed Attention, and DeepSeek-V4.1-Flash with CSA2, cross-layer cache reuse and FP4 caching bringing the global KV cache to 890 bytes per token, roughly 437x below DeepSeek-V1 — and claims Claude Opus 5.5 and GPT-6.1 Sol shipped with these techniques, with Opus 5.5 cutting cache-read pricing 60% versus Opus 5. The excerpt is truncated mid-sentence, and the pricing claims and model-release details are asserted by the author without cited primary sources.

Why: If the cache-read price cuts described here are real, the cost of running long-context coding and agent sessions shifts from output tokens toward a much cheaper cache-read line item, which changes how you'd budget and architect retrieval-heavy agents. But the article gives no links to DeepSeek's papers or to Anthropic/OpenAI pricing pages, so before repricing anything, verify the 890 bytes-per-token figure and the claimed 60% Opus cache-read reduction against the vendors' own docs — the HN thread (349 points, 368 comments) is a better starting point than the post itself.

Claire Vo recaps OpenAI DevDay 2026 from the floor and from her own early testing, covering ChatGPT Dots, Spaces, and Sites, GPT-6.1 Sol, a vision-capable Decisions API, Astra ultrafast, and updates to the Agents API, computer use, and plugins. Her hands-on demos include AI-picked podcast thumbnails, a collaborative sketchpad built on Astra ultrafast, and a prompt-driven 3D world her kids redesigned in real time — that last experiment cost about $97. The piece is framed as early impressions of what's promising, what still feels rough, and what to try first, not as benchmarks.

Why: The only hard number in the piece is a cost signal: one interactive 3D-world experiment on Astra ultrafast ran about $97, so if you're prototyping real-time or generative interactive apps, budget-test that pricing before promising it to a client or shipping it in a product. The two items worth a look for teams rather than solo demos are Spaces (human-agent collaboration) and Sites with connectors and plugins (sharing internal tools with scoped data permissions) — if you already expose internal tooling to agents, those permission semantics are the part to evaluate. Everything else here is a topic list; there are no latencies, version numbers, or API pricing in the text, so treat it as a triage list, not a technical evaluation.

OpenAI announced GPT-6.1 Sol, an upgrade to GPT-6 Sol that it claims nearly matches GPT-6 Astra on agentic coding, computer use and professional work at one-fifth of Astra's standard input/output token prices. Cached input is listed at $0.10 per million tokens, which OpenAI says is 95% below its standard input pricing and 50% below GPT-6 Sol's cached rate. The post cites self-reported results including matching GPT-6 Astra on DeepSWE v1.1 at roughly one-fifth the cost, beating GPT-6 Sol's best DeepSWE score by 6.4 percentage points at lower reasoning effort, and scoring 2.2 points above Opus 5.5 on AutomationBench at medium effort for about a third of the cost; the excerpt cuts off mid-sentence in the OSWorld 2.0 computer-use section, so those numbers are not visible here.

Why: The only decision-grade number in this post is cached input at $0.10 per million tokens, 50% below GPT-6 Sol's cached rate — if your agent loop resends the same system prompt, tool schemas or document context on every call, that is the line item that changes your bill, not the headline token price. Every capability claim (DeepSWE v1.1, GDP.pdf, AutomationBench) is OpenAI's own benchmark run with no independent replication, and the OSWorld 2.0 section is truncated, so treat this as a reason to re-run your own eval on one cached-context workload, not as a reason to migrate production traffic.