Compare·Provider
Continuum
Multi-agent workbench
VS
OpenCode
Open multi-provider coding agent

OpenCode in Continuum

Same OpenCode runtime when you pick it. Continuum is the multi-lab cockpit around it.

OpenCode is an open multi-provider coding agent (`opencode serve`, models.dev providers, auth.json). Continuum treats OpenCode as a first-class provider - and also hosts Claude Code, Codex, Cursor agent, and peers - with native clients, multi-account isolation for Claude/Codex, live gauges, and a local $ ledger. OpenCode’s models.dev provider catalog and auth.json are part of Continuum’s connector story (OpenRouter, Z.ai Coding, etc.). Continuum still adds Claude/Codex peers, mobile, gauges, and worktree Code panes OpenCode alone does not productize the same way.

Download for Mac All comparisons
Workflow Monday path Matrix Decide Money FAQ

Updated 2026-08-03 · Mac stable · Win/Linux desktop beta

Choose Continuum when

You want OpenCode plus Claude/Codex peers, Continuum mobile, worktree Code panes, multi-account gauges, and repo analytics - not only the OpenCode TUI/server alone.

Choose OpenCode when

You want OpenCode’s own CLI/server experience and are happy without Continuum’s multi-device workbench.

Snapshot provider · Continuum hosts it
Dimension Continuum OpenCode
Engine identity Hosts OpenCode + Claude + Codex… Is the OpenCode agent
Auth path opencode auth.json + lab CLIs auth.json / models.dev providers
Multi-account Claude/Codex instance isolation + OpenCode keys Provider keys in OpenCode store
Surfaces Continuum adds Chat · plan · diff · PR · terminal · artifacts · mobile clients OpenCode CLI / TUI / serve
Gauges Menu-bar 5h + weekly gauges Provider-specific / none Continuum-class
Analytics Local JSONL → $ by repo/provider/day Your API bills + OpenCode data
Mac stable · web · iPhone · Watch · Win/Linux desktop beta · free app
01 · OpenCode alone vs Continuum host
In Continuum

OpenCode in a multi-lab cockpit

10:00

Continuum session on OpenCode model for a BYOK path; second session on Claude Max for subscription rail.

13:00

Approve OpenCode plan-ready from iPhone. Diff pane shows the worktree.

17:00

Analytics columns show OpenCode/connector spend after history parse; Claude gauges show weekly heat. OpenRouter model for a cheap spike session; Claude Max for the production branch - same Continuum sidebar.

In OpenCode

OpenCode-native day

10:00

Run OpenCode CLI/serve directly. Pick models.dev providers from OpenCode’s own UX.

13:00

Stay in OpenCode’s surface. Continuum mobile and multi-lab sidebar are outside this path.

17:00

Cost is API keys + OpenCode workflow. No Continuum repo ledger unless you add Continuum. OpenCode serve alone is enough when Continuum clients and multi-lab gauges are out of scope.

02 · Monday path

Monday with OpenCode inside Continuum

Install, connect, first session - steps you can run the same day.

01

Install Continuum; ensure OpenCode is installed and auth.json has the providers you need.

02

Code tab → new session → OpenCode model from catalog (including connector vendors when configured).

03

Also spawn a Claude session if you want subscription-rail Claude beside OpenCode BYOK models.

04

Pair iPhone for plan/interrupt while OpenCode runs on the host.

05

Check analytics attribution for OpenCode/connector usage after DB parse cycles.

06

Keep `opencode` terminal for server debugging; Continuum for multi-device ops.

07

After adding an OpenRouter key in Continuum Settings, confirm the model appears in catalog before spawning.

How Continuum actually runs OpenCode

How Continuum actually runs OpenCode

  • OpenCode rides Continuum’s OpenCode process manager / serve path where integrated - models from ModelCatalog + probes.
  • Connector vendors (OpenRouter, Z.ai Coding) store keys in opencode auth.json; Continuum Settings writes the same store.
  • Code panes project session chat; mobile pair hits the Continuum daemon, not OpenCode’s UI alone.
  • Claude/Codex multi-account isolation remains separate (CLAUDE_CONFIG_DIR / CODEX_HOME) from OpenCode auth.
  • Analytics: OpenCode DB rows can attribute providerID; live SSE may roll under OpenCode until next parse - known nuance.
  • ProviderDescriptor ranks and Tahoe tokens keep OpenCode/connectors visually distinct in Usage and analytics when they own columns.
03 · OpenCode host matrix

Same engine under your login. Continuum cells name the cockpit - panes, gauges, isolation - not a re-score of model quality.

Capability Continuum OpenCode
OpenCode agent runtime First-class provider Is OpenCode
Claude Code peer in same app PTY host Not Claude Code product
Codex peer in same app ACP harness Not Codex product
Connector vendors (OpenRouter, Z.ai…) Via OpenCode auth + catalog models.dev providers
iPhone / Watch control Native iPhone + Watch · QR/Tailscale pair No Continuum pair
Worktree + PR panes Chat · plan · diff · PR · terminal · artifacts Agent + your git
Live multi-lab gauges Menu-bar 5h + weekly gauges Depends on provider
Local $ by repo Local JSONL → $ by repo/provider/day API invoices
Multi-account Claude Max CLAUDE_CONFIG_DIR / CODEX_HOME isolation Not Continuum multi-account
Free workbench app Free app · optional Plus/Max/Ultra hosted OSS/community + model costs
04 · What Continuum layers on
01 · Layering

Engine versus cockpit.

OpenCode is an agent runtime. Continuum is the multi-device workbench that can host OpenCode and peers. Not a hostile fork fight. Continuum ModelCatalog merges OpenCode/connector models into the Code tab picker with enable guards and live probes.

02 · Peers

OpenCode is not the only engine.

Continuum keeps Claude Code and Codex as first-class peers. OpenCode alone stays OpenCode-shaped by design. Continuum Settings can write OpenCode auth entries so keys do not live in a second ad-hoc file format.

03 · Connectors

Shared auth.json gravity.

Continuum Settings can write OpenCode auth for connector vendors - same store OpenCode uses. Continuum still owns mobile/gauges/worktrees.

04 · Ops

Phone, gauges, ledger.

Those are Continuum products. If you only need OpenCode TUI, stay on OpenCode.

05 · Coexistence
Stack recipe

Same engine. Bigger cockpit.

Stack recipe: Continuum can drive OpenCode sessions and also uses OpenCode’s auth.json path for connector vendors (OpenRouter, Z.ai Coding, etc.). You can still run `opencode` naked. Continuum does not replace OpenCode - it hosts it beside other engines.

Download Continuum Providers
06 · Host OpenCode or stay CLI-only
Continuum

Choose Continuum to operationalize OpenCode

  • Host OpenCode beside Claude Code and Codex in one sidebar. Continuum Code panes stay first-class: chat, plan, diff, PR, terminal, artifacts.
  • Need Continuum mobile clients and worktree Code panes.
  • Want multi-account Claude/Codex isolation plus OpenCode connectors.
  • Need live gauges and multi-lab local analytics.
  • Prefer free Continuum app as the daily cockpit.
OpenCode

Stay on raw OpenCode when the TUI is enough

  • Want only OpenCode’s CLI/serve UX. Continuum will not pretend to win that job with marketing wording alone.
  • Happy without Continuum mobile or multi-lab gauges.
  • Prefer pure OpenCode workflow for models.dev providers.
  • Do not need Claude Max multi-account stacking.
  • Want the lightest OpenCode-only install.
07 · What you pay
Continuum

App + your labs

Free app with BYOK / your subscriptions. Optional Continuum-hosted inference: Plus $25/mo, Max 100, Max 200, Ultra. OpenCode model usage bills your API keys / providers.

OpenCode

Their bill

OpenCode is typically free/OSS software; you pay model providers (OpenRouter, etc.). Verify OpenCode project + provider pricing as of 2026.

How to compare

Total cost of work

As of 2026 - verify on continuumcode.ai/pricing and the vendor site. Continuum does not mark up BYOK or subscription sessions it drives under your login. Continuum does not mark up your OpenCode provider keys.

Continuum ladder: Free app · Plus $25/mo · Max 100 · Max 200 · Ultra - full pricing. Also analytics, multi-account, devices.

08 · Questions

OpenCode + Continuum.

Deep dives: docs, providers, sessions.

Continuum hosts OpenCode and adds multi-device workbench features. If you only want OpenCode, use OpenCode. If you want Continuum ops around it (and peers), use Continuum.

No. Continuum integrates OpenCode as a provider path and shares auth.json for connectors.

Yes via OpenCode connector / Continuum provider setup that writes OpenCode auth - catalog and settings paths productize it.

Only if you want Claude sessions. OpenCode BYOK paths and Claude subscription rails are different engines Continuum can run side by side.

In OpenCode’s auth store (e.g. auth.json under the OpenCode data path). Continuum Settings can write entries; Continuum does not invent a second secret format for those connectors.

Live SSE events may not carry providerID the way DB rows do; Continuum’s analytics bridge attributes more precisely on history parse. That is a known nuance, not a silent surcharge.

No. Claude Code and Codex work without OpenCode. OpenCode matters when you want that runtime or connector vendors that ride auth.json.

Begin

Run OpenCode
from Continuum.

Free app. Your subscriptions. Optional hosted inference. Mac stable - Windows and Linux desktop are beta.

Download for Mac See pricing

vendor-neutral · local-first · multi-device