Function countTokensViaHaikuFallback

src/services/tokenEstimation.ts:257–332

(
  messages: Anthropic.Beta.Messages.BetaMessageParam[],
  tools: Anthropic.Beta.Messages.BetaToolUnion[],
)

 *   otherwise uses the configured SmallFastModel.
 */
export async function countTokensViaHaikuFallback(
  messages: Anthropic.Beta.Messages.BetaMessageParam[],
  tools: Anthropic.Beta.Messages.BetaToolUnion[],
): Promise<number | null> {
  // Check if messages contain thinking blocks
  const containsThinking = hasThinkingBlocks(messages)

  // If we're on Vertex and using global region, always use Sonnet since Haiku is not available there.
  const isVertexGlobalEndpoint =
    isEnvTruthy(process.env.CLAUDE_CODE_USE_VERTEX) &&
    getVertexRegionForModel(getSmallFastModel()) === 'global'
  // If we're on Bedrock with thinking blocks, use Sonnet since Haiku 3.5 doesn't support thinking
  const isBedrockWithThinking =
    isEnvTruthy(process.env.CLAUDE_CODE_USE_BEDROCK) && containsThinking
  // If we're on Vertex with thinking blocks, use Sonnet since Haiku 3.5 doesn't support thinking
  const isVertexWithThinking =
    isEnvTruthy(process.env.CLAUDE_CODE_USE_VERTEX) && containsThinking
  // On Noumena-managed first-party, route token counting through the main
  // loop model so the reported usage matches the tokenizer of the serving
  // model. GLM, Kimi, and DeepSeek V4 Flash each have provider-specific
  // tokenizers; using a different model family for counting would silently
  // mis-estimate input tokens, breaking auto-compact thresholds.
  const isManagedFirstParty = getAPIProvider() === 'firstParty'
  const model = isManagedFirstParty
    ? getMainLoopModel()
    : isVertexGlobalEndpoint || isBedrockWithThinking || isVertexWithThinking
      ? getDefaultFlashModel()
      : getSmallFastModel()
  const inferenceClient = await getInferenceClient({
    maxRetries: 1,
    model,
    source: 'count_tokens',
  })

  // Strip tool search-specific fields (caller, tool_reference) before sending
  // These fields are only valid with the tool search beta header
  const normalizedMessages = stripToolSearchFieldsFromMessages(messages)

  const messagesToSend: MessageParam[] =
    normalizedMessages.length > 0
      ? (normalizedMessages as MessageParam[])
      : [{ role: 'user', content: 'count' }]

  const betas = getModelBetas(model)
  // Filter betas for Vertex - some betas (like web-search) cause 400 errors
  // on certain Vertex endpoints. See issue #10789.
  const filteredBetas =
    getAPIProvider() === 'vertex'
      ? betas.filter(b => VERTEX_COUNT_TOKENS_ALLOWED_BETAS.has(b))
      : betas

  // biome-ignore lint/plugin: token counting needs specialized parameters (thinking, betas) that sideQuery doesn't support
  const response = await inferenceClient.createMessage({
    model: normalizeModelStringForAPI(model),
    max_tokens: containsThinking ? TOKEN_COUNT_MAX_TOKENS : 1,
    messages: messagesToSend,
    tools: tools.length > 0 ? tools : undefined,
    ...(filteredBetas.length > 0 && { betas: filteredBetas }),