Le Chat Pro's New Work Mode and Remote Agents: What You Actually Pay

$14.99/month is the price Mistral set to move you from chatbot to cloud-native agent. The gap between Free and Pro is wider than the headline number suggests, and the $24.99 Team tier has traps most teams don’t see coming until they hit the scalability wall.

What Work Mode Actually Does

Mistral’s May 2026 launch moved Le Chat past conversation into multi-step workflows. Work Mode lets agents handle inbox triage, draft replies to emails and messages, sync with your calendar, research across internal docs and the web, and create tickets in Jira or Sentry. All without babysitting every keystroke. The agent shows you every action before executing sensitive ones. This is the feature Mistral is selling as the Pro unlock.

The backbone is Medium 3.5, a 128B dense model with a 256k context window. Mistral’s launch announcement notes it runs self-hosted “on as few as four GPUs,” but Mistral is selling you the cloud hosting inside the $14.99 tag. InfoQ’s independent review confirmed the feature parity with Claude and ChatGPT’s agent modes—similar connector count, similar task scope, but Mistral’s integration layer into Le Chat ships native at no extra configuration cost.

Pro at $14.99 vs Free: The Real Feature Delta

The Free tier still exists, still has models, but hits you with soft caps. Mistral doesn’t publish the exact daily message ceiling for Free; their help docs say limits reset every three hours and “max daily limitations may apply.” You also get image generation, 40+ tool connectors, and 500 memory slots. That’s enough to treat Le Chat as Slack’s smarter sibling.

Pro ($14.99/month) removes the message soft cap. We tracked down the hard boundary on Mistral’s pricing page: Pro delivers “up to 6x Free” message throughput, which is how Mistral officially quantifies the gap. Work Mode turns on. Free gets you chatting; Pro gets you automation. Pro also includes 15GB of cloud storage (Mistral doesn’t publish the Free tier storage cap), no-telemetry mode where your messages don’t train Mistral’s next run, and the Vibe remote agent CLI hooks into Le Chat’s interface.

When Team at $24.99 Becomes Non-Optional

The Team plan is $24.99 per user per month (or $19.99 annually), and it’s not a free-tier-to-shared-workspace upgrade. It’s built for orgs that want governance or scale.

You get 30GB per user (Pro gets 15GB), shared RAG libraries so your team can pool knowledge bases, admin controls to enforce data-sharing policies, and team-level audit logs. TechJack’s 2026 matrix notes that Team accounts are opted out of data sharing by default. Mistral doesn’t use your work to train, period.

The trap: per-seat pricing means a five-person team is $149.95/month and a ten-person team is $299.90. There’s no named per-message overage fee, but Team doesn’t include higher absolute message limits. If your team hits a bandwidth wall and every seat maxes out at the Pro limit simultaneously, Mistral doesn’t publicly detail the next level until you contact Enterprise sales.

API Costs if You Build Custom Agents

Not everyone runs agents inside Le Chat. If you’re building on the API, Medium 3.5 costs $1.50 per million input tokens and $7.50 per million output tokens. For a typical 5K-token request with a 2K response, that works out to $0.0225 total. Compare that to Claude Sonnet, which runs $3.00 input and $15.00 output per million tokens per the Anthropic API docs. The same 5K-in/2K-out call on Sonnet costs $0.045, roughly twice as much. Run 1,000 Medium 3.5 queries a month and you’re at $22.50 in API costs. That’s more expensive than the $14.99 Pro subscription if your usage is light, so the hosted Le Chat plan wins for solo usage. API billing gets aggressive once you’re spinning up agents that iterate.

What Mistral Still Won’t Tell You

Work Mode integrations are live for email, calendar, Jira, GitHub, and Sentry. Mistral has not published per-integration rate limits, execution timeouts, or whether concurrent agents count toward message quotas. If your team spins up five agents and they all hit the API simultaneously, does that burn five messages or one? Mistral’s docs are silent. Enterprise customers will get these answers under NDA; the rest of us are flying blind.

Also unpublished: whether Free and Pro accounts can enable Work Mode on the same API key, or if each seat needs its own subscription. Team accounts get shared access, but the granularity of “user A runs agents, user B reads outputs” hasn’t been clarified.

Our Take

Solo developers and small teams (under 5 people): Start on Free, upgrade to Pro the moment you want to automate a workflow. At $14.99/month it’s cheaper than Claude Pro and the Work Mode connectors for calendar, Jira, and GitHub ship natively—ChatGPT Plus ($20/month) requires custom GPT configuration to get the same workflows.

Teams of 5 to 20: The Team plan is the right move if you’re pooling knowledge bases or need audit logs for compliance. Lock in annual billing ($19.99/user/month) or the per-seat cost blows past tighter budgets. Don’t pay month-to-month unless churn is expected. We’d compare the per-seat math against what similar teams are spending on Claude Max versus ChatGPT Pro before committing.

Enterprise and heavily integrated setups: Ask Mistral for pricing. Custom models, on-prem options, and white-label UI exist, but the asking price starts well above what they’ll advertise.

The biggest play Mistral made here isn’t the agent features themselves. It’s pricing the Free tier so that Free users look at Pro and say “that’s a reasonable $180 a year for my side project.” That’s a psychology win. The real question is whether Work Mode’s integrations stay permissionless or whether Mistral starts hiding premium connectors behind higher tiers. Watch the June updates closely. If you want to see how Medium 3.5’s API costs stack up against the full field, we ran those numbers in our LLM API price war breakdown.