Tag
#podcast
From News
News · 2026-11-12
Google puts Gemini Decision 1 on fast choices where a large model is overkill
Google introduced Gemini Decision 1 for closed choices with up to 255 options, 256K context and a price of $0.10 per million input tokens. Independent measurements cited by Latent Space show that a fast specialist does not automatically deliver the best decisions or the lowest cost per task.
Read →News · 2026-11-10
Resend wins agent recommendations by designing its interface for machines too
Claude Code chose Resend first in 64 of 102 extractable email tool recommendations. This is not market share, but it is a practical demonstration that agent-readable documentation is becoming a distribution channel for developer products.
Read →News · 2026-11-10
Fable 5.2 cut token prices, but heavy reasoning can raise the bill
Anthropic released Claude Fable 5.2 at $8 per million input tokens and $40 per million output tokens. Independent tests place it near the top, while exposing a crucial paradox: cheaper tokens can produce a more expensive task when the model verifies more aggressively.
Read →News · 2026-11-08
Stanford changes the assignment: students must defend agent written code
In the second run of Stanford's CS146S, students work in an unfamiliar open source repository and explain their change to its maintainer. Assessment moves from producing code toward understanding, context and responsibility.
Read →News · 2026-11-06
Inkling 2 Small trained in 68 hours, but its open weights still await safety tests
Thinking Machines Lab has introduced Inkling 2 Small, a multimodal MoE model with 594 billion total parameters and 17 billion active parameters. Training took 68 hours on fewer than 12 trillion tokens, but full weights will follow only after additional safety testing.
Read →News · 2026-11-03
Astra edited 0.016% of Qwen's weights and lifted MATH-500 from 7.9% to 48.1%
MIT researchers let a frontier model directly inspect and modify small neural networks without gradient training. The headline result fixes a repetitive reasoning loop, making it a targeted failure repair rather than automatic acquisition of mathematical skill.
Read →News · 2026-10-31
Arca keeps financial agents on 10-day jobs but keeps money movement human
Financial firm Arca says its agents coordinate client work that can span up to 10 days through persistent memory. The practical lesson is not longer model runtime, but durable state, a custom harness and a hard boundary before money moves.
Read →News · 2026-10-27
Anthropic rewrote the Messages API in Rust and put credentials out of its agents' reach
The heads of engineering and product for Claude Platform told Latent Space how Anthropic used Claude to port the Messages API from Python to Rust, saving an estimated 40 to 50 % of CPU. For teams building agents, the more useful part is why Managed Agents keep credentials outside the sandbox.
Read →News · 2026-10-12
Dynamic workflows move the agent plan from context into executable code
Anthropic has made dynamic workflows available, letting Claude write JavaScript that coordinates tens to hundreds of subagents and can be run again. The central change is not the agent count but moving orchestration outside one model's context window.
Read →News · 2026-10-10
Standard Bots Proves Factory Automation Hinges on Edge Data, Not Massive Frontier Models
US industrial robot maker Standard Bots raised $200 million at a $1 billion valuation. Rather than chasing hundred-billion-parameter models, the company deploys nimble low-billion-parameter architectures tuned via a few dozen on-site operator corrections.
Read →News · 2026-10-07
Stacklok moves agents from laptops to Kubernetes, along with enterprise control
Stacklok is building the open source Mecatl agent harness to separate the agent loop from the model, state and tool execution. The Kubernetes co-creators are betting that enterprises want central identity, policy and audit instead of agents scattered across laptops.
Read →News · 2026-10-02
AI writes 60% of Airbnb code, but handoffs are the bigger change
Airbnb CTO Ahmad Al-Dahle says AI authors 60% of company code and average pull request throughput per engineer is up 1.6 times. The deeper change is a move from documents and handoffs to prototypes, the Everest context layer and use-case-specific evals.
Read →News · 2026-09-25
Runway generates an interactive world at 24 fps, but without hard state
Runway's GWM Worlds 2 research preview streams interactive 720p video at 24 fps with 48,000 Hz audio. WorldPrompt separates persistent world context from timed actions, yet remains a prompt format without scripting or reliable state.
Read →News · 2026-09-23
Opus 5.5 cuts token prices 20%, but the bill per task may not fall
Anthropic launched Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens, 20% below Opus 5. Anthropic claims 40% lower costs on typical workloads, while an independent max-effort measurement shows an almost unchanged cost per task because the model uses more tokens.
Read →News · 2026-07-25
Opus 5 narrowly matched Fable 5, but benchmarks can no longer measure reality
The recent launch of Claude Opus 5 is accompanied by a confused community reaction. Epoch measured an ECI of 159 for the model versus 161 for Fable 5, but users report a much stronger practical feel for agentic behavior.
Read →News · 2026-09-15
Startup trains AI agents on train games to figure out how to get financial analysis skills out of them
Good Start Labs uses 19th-century train simulators to train artificial intelligence. According to Latent Space, they discovered that how the game is structured determines whether the agent transfers those skills to real-world business research.
Read →News · 2026-09-15
Everyone Wants to Audit AI: OpenAI, Anthropic, and xAI Sign AEF-1
Frontier labs agree on a common standard for third-party auditors for the first time. Instead of state regulators, companies are choosing their own embedded teams with unlimited access to training runs.
Read →News · 2026-09-09
OpenAI solved the Navier-Stokes equation with brute force for $40 million
OpenAI used 10,000 agents and 130 billion tokens to discover a singularity in the Navier-Stokes equations. Instead of elegant math, massive computing power won.
Read →News · 2026-09-07
Astra Dominates New AEO Tracker, While Anthropic Prefers Broader Research
The new AEO Tracker by Latent Space tested 7 models across 161 categories. Astra wins with fewer sources and higher assertiveness, while Claude bets on extensive search.
Read →News · 2026-09-01
A Halt to Community PRs: Major AI Projects Replace Humans with Their Own Agents
Top open source projects like Vercel AI SDK and tldraw are stopping external pull requests. They are replacing them with so-called software factories, where errors are reproduced, fixed, and reviewed by a fleet of their own specialized agents.
Read →News · 2026-09-01
Fal Pushes Video Latency Below the Real-Time Threshold
Fal's new H3 Max Live service can generate AI video faster than a viewer can watch it. The startup took an existing model from Minimax, sped it up 35x, and hooked it up to Twitch chat for an infinite interactive broadcast.
Read →News · 2026-08-22
The evolution of agent frameworks: the interface no longer guards the model, but our attention
Latent Space analyzes the shift in agent architectures. What used to be done by external software wrappers is being absorbed directly into the model weights. The remaining infrastructure is changing its purpose – it no longer controls AI, but filters our interaction with it.
Read →News · 2026-07-21
Xaira bets on X-Atlas because bigger models hit a wall without causal data
Latent Space profiles Xaira Therapeutics and its X-Cell model for drug discovery. The sober point is that cell models cannot be rescued by more parameters if the data lacks causal interventions.
Read →News · 2026-07-24
FLUX 3 Video pushes Black Forest Labs into multimodal production
Latent Space describes FLUX 3 as a multimodal flow model for video, audio, keyframes and longer sequences. If the performance claims and the open Dev version hold, Black Forest Labs is no longer just an image lab, but a supplier of a generative production layer.
Read →News · 2026-07-23
Poolside frames the model factory as the product, not just one model
Latent Space’s interview with Eiso Kant presents Poolside as a model factory: fewer than 70 researchers, 10000 to 20000 experiments per month and Laguna S 2.1 with 118B parameters. The story is about industrializing training, not just another coding model.
Read →News · 2026-06-15
Microsoft used Build to act like a model lab, not just a distributor
Latent Space frames Microsoft Build as the moment Microsoft showed its own MAI models alongside Copilot, Windows and Web IQ. The key ambition is to control data, inference and developer workflow at once, rather than leaving that leverage to partners.
Read →News · 2026-06-15
Andon Labs tests agents where benchmarks stop: money, people and shelves
Latent Space's interview with Andon Labs shows evals that look less like exams and more like running a small business. The key ingredients are long horizons and real consequences.
Read →News · 2026-06-15
Bad RL environments do not train agents, they teach them to trust a broken world
Latent Space published Auriel W's piece on why low-quality RL environments damage agent training. The point is simple: in reinforcement learning, the environment is the data generator, so a harness bug becomes training material.
Read →News · 2026-06-09
Agent cost is no longer a footnote. It is an engineering expense
Simon Willison shows how he manually added pricing for Claude Fable 5 in AgentsView and immediately saw the cost of local coding agents by project. The small trick points to a bigger shift: AI coding is starting to look like infrastructure consumption, not an app subscription.
Read →News · 2026-06-02
GitHub is preparing for a world where agents write commits at scale
The Latent Space interview with Kyle Daigle frames GitHub as a platform under pressure from agentic coding. The point is not another Copilot feature, but whether infrastructure built for human pace can absorb software produced by machines.
Read →