AI NewsSeptember 12, 2026

OpenAI launched a Data agent for ChatGPT Work that connects to company databases, plus ChatGPT for Financial Services, built with Morgan Stanley and Evercore for investment banking research and pitchbooks.
OpenAI introduced two ChatGPT Work tools on September 10. A Data agent connects to sources like BigQuery, Snowflake, and Datadog, turning questions into dashboards. ChatGPT for Financial Services, built with Morgan Stanley and Evercore, pairs GPT-6 Astra with data from PitchBook and Crunchbase.
Both target work usually done by junior staff: pulling numbers and building the pitchbooks that take analysts hours. Every output cites its source, so decisions stay traceable as more research gets automated.
Key Capabilities:

• Plain-language analytics: The Data agent turns questions into shareable dashboards, no queries or new tools needed.
• Wall Street-ready: Built with Morgan Stanley and Evercore for LBO modeling, buyer screening, and pitchbooks.
• Source-linked: Every figure traces back to its underlying data, from company databases to PitchBook filings.

Meta's Muse Spark 1.3, released September 2, reportedly jumped on the independent Agent Arena leaderboard this week and scores 61 on Artificial Analysis's Intelligence Index, tying top rival models at a lower cost.
Meta released Muse Spark 1.3 on September 2. It reportedly jumped on the independent Agent Arena leaderboard this week and scores 61 on Artificial Analysis's Intelligence Index, tying GPT-5.6 Sol and Grok 4.6.
That ties top models at a lower cost, and Meta says it asks more clarifying questions while cutting unneeded tool calls by 20%. On Meta's own chart it still loses most head-to-head agent tests to Claude Opus 5.
Key Capabilities:

• Efficiency gains: Uses 25% fewer tokens and cuts unnecessary tool calls by 20%, per Meta.
• Priced to compete: Available via Muse Code subscriptions starting at $5 a month.
• Mixed picture: Wins coding and long-context benchmarks but loses most agent tasks to Claude Opus 5.

DeepSeek's V4.1 Flash processed 1 trillion tokens on OpenRouter within 24 hours of launch, putting it on pace for the biggest 48-hour debut of any paid model tracked.
DeepSeek released V4.1 Flash on September 10, and it hit 1 trillion tokens on OpenRouter in 24 hours. That pace puts it on track for 2.8 trillion tokens by 48 hours, the biggest launch of any paid model.
Most of those tokens are cheap cache reads, priced around $0.006 per million, about five times cheaper than comparable models. That kind of volume signals developers are already routing real workloads to it, not just testing it.
Key Capabilities:

• Explosive demand: 1 trillion tokens processed in 24 hours, nearly double by hour 36.
• Cheap by design: 90% of usage is cache reads, priced at just $0.006 per million tokens.
• Built for agents: DeepSeek says it beats its own V4 Pro on speed and task completion.

Black Forest Labs launched FLUX Video Edit, letting anyone alter an existing AI video by removing objects, changing colors, or swapping dialogue, undercutting rivals on both speed and price.
Black Forest Labs launched FLUX Video Edit on September 10, calling it the fastest, cheapest video editing model available. It edits an existing clip: adding, removing, or replacing objects and characters, changing backgrounds, or editing dialogue with lip sync.
It changes only what you ask for and leaves the rest of the footage untouched. At $0.03 per second, that puts precise AI video editing within reach of everyday creators, not just big studios.
Key Capabilities:

• Surgical edits: Add, remove, or replace objects and characters while the rest of the frame stays untouched.
• Dialogue swaps: Change or translate spoken lines with accurate lip sync in the edited clip.
• Rock-bottom pricing: A 10-second edit costs about 30 cents, undercutting rivals like Omni and Seedance.

Suno launched v6, its first AI music models built with licensed music from Warner Music, BMG, and Believe, while ElevenLabs shipped Music v2.5 as its new default song generator with richer melodies and lifelike instruments.
Suno released v6 on September 9, its first models built on licensed music from Warner, BMG, and Believe. ElevenLabs followed with Music v2.5, its new default song generator with richer melodies and lifelike instruments.
Both mark AI music's shift from lawsuits to licensing. Suno still faces suits from Universal and Sony, but paying Warner and BMG shows labels can work with AI, not just sue it. That decides who profits from the next AI hit.
Key Capabilities:

• Licensed foundation: Suno v6 trains on music from Warner, BMG, and Believe, with revenue sharing for rights holders.
• Free tier upgrade: ElevenLabs Music v2.5 gives free-plan users commercial rights to what they create.
• Plain-language editing: Both let users rewrite a single lyric or section without technical prompts.